AMD EPYC 7002 / 7003 series NAS (512 GB) spec sheet
What the AMD EPYC 7002 / 7003 series NAS (512 GB) can run
We check your GPU or device against every model. Browsers can’t see how much memory you have, so confirm your exact model below.
- Unified memory
- 512 GB
- Usable for models
- 471 GB
- Memory bandwidth
- 205 GB/s
- FP16 compute
- 1.0 TFLOPS
Open in Model Finder Best LLMs by task Compare with another GPU
LLMs on the AMD EPYC 7002 / 7003 series NAS (512 GB)
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 96Excellent quality (A+) · 96 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax M3.1 Flash (428B-A23B) | Q6_K | 391 GB | 4.2 tok/s | Slow |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | Ling 3.1 Flash 1019 (560B-A25B) | Q4_K_M | 387 GB | 5.5 tok/s | Slow |
| 93Excellent quality (A+) · 93 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | K2-Horizon (375B-A23B) | Q6_K | 351 GB | 4.2 tok/s | Slow |
| 93Excellent quality (A+) · 93 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax M3 (427B) | Q6_K | 389 GB | 3.8 tok/s | Slow |
| 91Excellent quality (A+) · 91 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | Beam 1.0 (501B-A23B) | Q5_K_M | 394 GB | 4.8 tok/s | Slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | Ling 3.1 Flash (560B-A25B) | Q4_K_M | 387 GB | 5.5 tok/s | Slow |
| 94Excellent quality (A+) · 94 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | Step 5 Preview (600B-A27B) | Q5_K_M | 468 GB | 4.2 tok/s | Slowtight fit |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.7 (355B-A32B) | Q5_K_M | 296 GB | 3.6 tok/s | Slow |
| 91Excellent quality (A+) · 91 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.8 Flash Next (125B-A6B) | Q4_K_M | 148 GB | 21.6 tok/s | Runs well |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Coder 480B-A35B | Q5_K_M | 381 GB | 3.3 tok/s | Slow |
| 82Great quality (A) · 82 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V4.1 Flash 1017 (552B-A16B) | Q3_K_M | 412 GB | 10.9 tok/s | Runs welltight fit |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V4 Flash (284B-A13B) | Q4_K_M | 224 GB | 10.4 tok/s | Runs well |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax M2.7 (230B-A10B) | Q5_K_M | 203 GB | 11.1 tok/s | Runs well |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V3.2 (671B) | Q4_K_M | 457 GB | 3.7 tok/s | Slowtight fit |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 0528 (671B) | Q4_K_M | 456 GB | 3.7 tok/s | Slowtight fit |
| 81Great quality (A) · 81 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V4.1 Flash (552B-A16B) | Q3_K_M | 412 GB | 10.9 tok/s | Runs welltight fit |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax M2.5 (230B-A10B) | Q5_K_M | 203 GB | 11.1 tok/s | Runs well |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V3.2 Exp | Q4_K_M | 457 GB | 3.7 tok/s | Slowtight fit |
| 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | LongCat Flash Chat (560B) | Q4_K_M | 381 GB | 5.1 tok/s | Slow |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.6 35B-A3B | Q4_K_M | 60.4 GB | 40.2 tok/s | Runs great |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 (671B) | Q4_K_M | 448 GB | 3.7 tok/s | Slowtight fit |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Inkling Small (276B-A12B) | Q4_K_M | 201 GB | 11.2 tok/s | Runs well |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM-V 4.7 35B-A3B | Q4_K_M | 59.9 GB | 40.2 tok/s | Runs great |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 35B-A3B | Q4_K_M | 60.4 GB | 40.2 tok/s | Runs great |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V3.1 Terminus | Q4_K_M | 456 GB | 3.7 tok/s | Slowtight fit |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V4 Flash (April 2026) | Q4_K_M | 216 GB | 10.4 tok/s | Runs well |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax M2.1 (230B-A10B) | Q5_K_M | 203 GB | 11.1 tok/s | Runs well |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Flint (40B-A2.5B) | Q5_K_M | 66.8 GB | 39.7 tok/s | Runs great |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V3.1 (671B) | Q4_K_M | 456 GB | 3.7 tok/s | Slowtight fit |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Nex N2 Mini (35B-A3B) | Q4_K_M | 59.9 GB | 40.2 tok/s | Runs great |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Kolibri 1 (78B-A3.5B) | Q4_K_M | 86.1 GB | 35.5 tok/s | Runs great |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Step 3.7 Flash (198B-A11B) | Q5_K_M | 182 GB | 10.1 tok/s | Runs well |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Sage-A1.5 35B-A3B | Q4_K_M | 60.4 GB | 40.2 tok/s | Runs great |
| 85Excellent quality (A+) · 85 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Large 3 (675B) | Q4_K_M | 451 GB | 3.4 tok/s | Slowtight fit |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | KAT-Coder V2.5 Dev (35B-A3B) | Q4_K_M | 59.6 GB | 40.2 tok/s | Runs great |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax M2 (230B-A10B) | Q5_K_M | 203 GB | 11.1 tok/s | Runs well |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Step 3.5 Flash (196B-A11B) | Q5_K_M | 180 GB | 10.1 tok/s | Runs well |
| 85Excellent quality (A+) · 85 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Ornith 1.5 35B-A3B | Q4_K_M | 60.4 GB | 40.2 tok/s | Runs great |
| 82Great quality (A) · 82 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | GLM-5.3 (753B-A40B) | Q3_K_M | 408 GB | 4.4 tok/s | Slowtight fit |
| 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-Next 80B-A3B Thinking | Q4_K_M | 88.1 GB | 40.2 tok/s | Runs great |
| 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Coder Next (80B-A3B) | Q4_K_M | 87.1 GB | 40.2 tok/s | Runs great |
| 84Great quality (A) · 84 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Apodex 1.1 Mini (35B-A3B) | Q4_K_M | 60.4 GB | 40.2 tok/s | Runs great |
| 84Great quality (A) · 84 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Ornith 1.0 35B-A3B | Q4_K_M | 60.4 GB | 40.2 tok/s | Runs great |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Trinity Large Thinking (400B-A13B) | Q4_K_M | 282 GB | 10.4 tok/s | Runs well |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 4 26B-A4B | Q4_K_M | 54.4 GB | 32.7 tok/s | Runs great |
| 81Great quality (A) · 81 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.6 (355B-A32B) | Q5_K_M | 295 GB | 3.6 tok/s | Slow |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | K-EXAONE 236B-A23B | Q6_K | 233 GB | 4.2 tok/s | Slow |
| 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.7 Flash (30B-A3B) | Q4_K_M | 57.8 GB | 40.2 tok/s | Runs great |
| 80Great quality (A) · 80 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | GLM-5.2 (753B-A40B) | Q3_K_M | 408 GB | 4.4 tok/s | Slowtight fit |
| 80Great quality (A) · 80 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Muse Spark Open (816B-A42B) | Q3_K_M | 442 GB | 4.2 tok/s | Slowtight fit |
| 79Great quality (A) · 79 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Next 80B-A3B Instruct | Q4_K_M | 88.1 GB | 40.2 tok/s | Runs great |
| 84Great quality (A) · 84 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | gpt-oss 120B | MXFP4 | 105 GB | 26.7 tok/s | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Intern-S2 (397B-A17B) | Q3_K_M | 236 GB | 10.3 tok/s | Runs well |
| 82Great quality (A) · 82 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Agents-A1 35B-A3B | Q4_K_M | 59.9 GB | 40.2 tok/s | Runs great |
| 82Great quality (A) · 82 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Qwen AgentWorld 35B-A3B | Q4_K_M | 59.6 GB | 40.2 tok/s | Runs great |
| 79Great quality (A) · 79 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 397B-A17B | Q3_K_M | 236 GB | 10.3 tok/s | Runs well |
| 79Great quality (A) · 79 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Nex N2 Pro (397B-A17B) | Q3_K_M | 233 GB | 10.3 tok/s | Runs well |
| 79Great quality (A) · 79 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | GLM-5.1 (744B-A40B) | Q3_K_M | 408 GB | 4.4 tok/s | Slowtight fit |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 Nemotron Ultra 253B v1 | Q8_0 | 311 GB | 0.3 tok/s | Very slow |
| 79Great quality (A) · 79 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | GLM-5 (744B-A40B) | Q3_K_M | 408 GB | 4.4 tok/s | Slowtight fit |
| 79Great quality (A) · 79 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.5 (355B-A32B) | Q5_K_M | 296 GB | 3.6 tok/s | Slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Ornith 1.5 397B-A17B | Q3_K_M | 236 GB | 10.3 tok/s | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan Hy4 Preview (780B) | Q3_K_M | 421 GB | 3.2 tok/s | Slowtight fit |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Ornith 1.5 122B-A10B | Q5_K_M | 127 GB | 11.1 tok/s | Runs well |
| 78Great quality (A) · 78 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax-M1 80K | Q4_K_M | 317 GB | 3.0 tok/s | Slow |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Command A+ (218B-A25B) | Q6_K | 218 GB | 3.8 tok/s | Slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3 Ultra 550B-A55B | Q3_K_M | 313 GB | 3.2 tok/s | Slow |
| 85Excellent quality (A+) · 85 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3.5 241B-A28B | Q6_K | 237 GB | 3.4 tok/s | Slow |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 122B-A10B | Q5_K_M | 127 GB | 11.1 tok/s | Runs well |
| 77Great quality (A) · 77 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax-M1 40K | Q4_K_M | 317 GB | 3.0 tok/s | Slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan Hy-Memory 30B-A3B | Q4_K_M | 57.4 GB | 40.2 tok/s | Runs great |
| 78Great quality (A) · 78 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V3 0324 (671B) | Q4_K_M | 456 GB | 3.7 tok/s | Slowtight fit |
| 81Great quality (A) · 81 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | MiMo V2.6 Flash (310B-A15B) | Q3_K_M | 191 GB | 11.6 tok/s | Runs well |
| 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3.5 Bolt 74B-A9B | Q6_K | 99.8 GB | 10.4 tok/s | Runs well |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3.5 Super VL 120B-A12B | Q4_K_M | 114 GB | 11.2 tok/s | Runs well |
| 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | Raon-1 mini Preview (30B-A4B) | Q4_K_M | 55.9 GB | 29.3 tok/s | Runs well |
| 79Great quality (A) · 79 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | Naive N0.5 Flash (309B-A15.5B) | Q3_K_M | 190 GB | 11.2 tok/s | Runs well |
| 77Great quality (A) · 77 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | Laguna XS 2.1 (33B-A3B) | Q4_K_M | 60.0 GB | 40.2 tok/s | Runs great |
| 78Great quality (A) · 78 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | MiMo V2.5 (310B-A15B) | Q3_K_M | 191 GB | 11.6 tok/s | Runs well |
| 77Great quality (A) · 77 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | Ling 3.0 Flash (127B-A5B) | Q4_K_M | 116 GB | 25.1 tok/s | Runs well |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | North Mini Code 1.0 (30B-A3B) | Q4_K_M | 57.2 GB | 40.2 tok/s | Runs great |
| 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Laguna S 2.1 (118B-A8B) | Q4_K_M | 112 GB | 16.5 tok/s | Runs well |
| 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 4 405B | Q6_K | 375 GB | 0.2 tok/s | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | MiMo V2 Flash (309B-A15B) | Q3_K_M | 190 GB | 11.6 tok/s | Runs well |
| 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax-Text-01 | Q4_K_M | 317 GB | 3.0 tok/s | Slow |
| 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3 Super 120B-A12B | Q4_K_M | 114 GB | 11.2 tok/s | Runs well |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron Cascade 2 30B-A3B | Q4_K_M | 57.6 GB | 35.2 tok/s | Runs great |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Small 4 (119B-A6B) | Q4_K_M | 111 GB | 21.6 tok/s | Runs well |
| 71Great quality (A) · 71 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V3 (671B) | Q4_K_M | 448 GB | 3.7 tok/s | Slowtight fit |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 30B-A3B Thinking 2507 | Q4_K_M | 57.7 GB | 37.0 tok/s | Runs great |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Coder 30B-A3B | Q4_K_M | 57.7 GB | 37.0 tok/s | Runs great |
| 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K2.6 (1T) | Q2_K | 450 GB | 7.0 tok/s | Slowtight fit |
| 73Great quality (A) · 73 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 30B-A3B Instruct 2507 | Q4_K_M | 57.7 GB | 37.0 tok/s | Runs great |
| 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | MiMo V2.6 Pro (1T-A42B) | Q2_K | 449 GB | 5.4 tok/s | Slowtight fit |
| 75Great quality (A) · 75 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Solar Open 2 250B | Q3_K_M | 163 GB | 11.6 tok/s | Runs well |
| 71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 24B-A2B | Q5_K_M | 55.5 GB | 40.4 tok/s | Runs great |
| 75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Tülu 3 405B | Q6_K | 375 GB | 0.2 tok/s | Very slow |
| 94Excellent quality (A+) · 94 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.8 27B | Q6_K | 61.6 GB | 3.5 tok/s | Slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K2.5 (1T) | Q2_K | 450 GB | 7.0 tok/s | Slowtight fit |
| 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K2.7 Code (1T) | Q2_K | 450 GB | 7.0 tok/s | Slowtight fit |
| 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Kimi Linear 48B-A3B | Q4_K_M | 68.3 GB | 40.2 tok/s | Runs great |
| 73Great quality (A) · 73 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3 Nano 30B-A3B | Q4_K_M | 57.6 GB | 35.2 tok/s | Runs great |
| 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 30B-A3B Thinking | Q4_K_M | 58.0 GB | 37.0 tok/s | Runs great |
| 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | MiMo V2.5 Pro (1T-A42B) | Q2_K | 448 GB | 5.4 tok/s | Slowtight fit |
| 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K2 Thinking (1T) | Q2_K | 449 GB | 7.0 tok/s | Slowtight fit |
| 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Maple Preview (20B-A1B) | Q8_0 | 59.8 GB | 40.4 tok/s | Runs great |
| 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Large 4 (1T-A52B) | Q2_K | 460 GB | 4.4 tok/s | Slowtight fit |
| 93Excellent quality (A+) · 93 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.6 27B | Q6_K | 61.6 GB | 3.5 tok/s | Slow |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Ling 2.6 Flash (104B-A7.4B) | Q4_K_M | 104 GB | 17.7 tok/s | Runs well |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Nanbeige4.3 33B | Q5_K_M | 63.5 GB | 3.5 tok/s | Slow |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | DiffusionGemma 26B-A4B IT | Q4_K_M | 54.4 GB | 31.2 tok/s | Runs great |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Ring Flash 2.0 (100B-A6B) | Q4_K_M | 102 GB | 21.3 tok/s | Runs well |
| 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Inkling (975B-A41B) | Q2_K | 420 GB | 5.5 tok/s | Slowtight fit |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Devstral 2 (123B) | Q8_0 | 174 GB | 0.6 tok/s | Very slow |
| 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K2 Instruct 0905 (1T) | Q2_K | 449 GB | 7.0 tok/s | Slowtight fit |
| 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 30B-A3B Instruct | Q4_K_M | 58.0 GB | 37.0 tok/s | Runs great |
| 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 30B-A3B | Q4_K_M | 57.7 GB | 37.0 tok/s | Runs great |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 27B | Q6_K | 61.6 GB | 3.5 tok/s | Slow |
| 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3.5 Lightning 30B-A3B | Q4_K_M | 57.6 GB | 40.2 tok/s | Runs great |
| 78Great quality (A) · 78 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.6V | Q4_K_M | 106 GB | 11.2 tok/s | Runs well |
| 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Jamba 1.5 Large | Q6_K | 365 GB | 1.0 tok/s | Very slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Sarvam 30B | Q5_K_M | 61.3 GB | 41.2 tok/s | Runs great |
| 65Good quality (B) · 65 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | NorthCUA Pro | Q3_K_M | 233 GB | 10.3 tok/s | Runs well |
| 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Ling Flash 2.0 (100B-A6B) | Q4_K_M | 102 GB | 21.3 tok/s | Runs well |
| 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Muse Glimmer 30B | Q6_K | 62.8 GB | 3.3 tok/s | Slow |
| 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K2 Instruct (July 2025) | Q2_K | 449 GB | 7.0 tok/s | Slowtight fit |
| 70Great quality (A) · 70 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | gpt-oss 20B | MXFP4 | 50.4 GB | 36.4 tok/s | Runs great |
| 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 405B Instruct | Q6_K | 375 GB | 0.2 tok/s | Very slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 4 31B | Q6_K | 65.5 GB | 3.1 tok/s | Slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Mellum2.1 Thinking (12B-A2.5B) | Q5_K_M | 47.0 GB | 39.7 tok/s | Runs great |
| 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Leanstral 119B-A6B | Q4_K_M | 111 GB | 20.0 tok/s | Runs well |
| 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Ling 2.6 1T | Q2_K | 449 GB | 3.6 tok/s | Slowtight fit |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | TMax 27B | Q6_K | 61.3 GB | 3.5 tok/s | Slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE 4.5 21B-A3B Thinking | Q4_K_M | 52.1 GB | 40.2 tok/s | Runs great |
| 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Ling 1T | Q2_K | 441 GB | 4.5 tok/s | Slowtight fit |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-Omni 30B-A3B Instruct | Q4_K_M | 60.6 GB | 37.0 tok/s | Runs great |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.5V | Q4_K_M | 106 GB | 11.2 tok/s | Runs well |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Solar Open 100B | Q4_K_M | 103 GB | 11.2 tok/s | Runs well |
| 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | GLM-5.3 Flash (320B-A18B) | Q2_K | 167 GB | 12.3 tok/s | Runs well |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Fara 1.5 27B | Q6_K | 61.3 GB | 3.5 tok/s | Slow |
| 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Ling 3.0 flash Sante | Q4_K_M | 116 GB | 25.1 tok/s | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Medium 3.5 (128B) | Q8_0 | 177 GB | 0.6 tok/s | Very slow |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.5 Air (106B-A12B) | Q4_K_M | 107 GB | 11.2 tok/s | Runs well |
| 82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Command A Reasoning (111B) | Q8_0 | 157 GB | 0.6 tok/s | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Llama 4 Maverick (17B x 128E) | Q3_K_M | 237 GB | 10.3 tok/s | Runs well |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Groot 65B A6B 261023 | Q4_K_M | 111 GB | 21.6 tok/s | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Large 2.1 (2411) | Q8_0 | 171 GB | 0.6 tok/s | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Trinity Mini (26B-A3B) | Q4_K_M | 54.4 GB | 40.2 tok/s | Runs great |
| 78Great quality (A) · 78 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 4 12B | Q4_K_M | 46.2 GB | 11.7 tok/s | Runs well |
| 71Great quality (A) · 71 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Sarvam 105B | Q4_K_M | 103 GB | 13.3 tok/s | Runs well |
| 85Excellent quality (A+) · 85 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | UI-Mate 27B | Q6_K | 61.3 GB | 3.5 tok/s | Slow |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | K2 Horizon 7B | Q4_K_M | 45.0 GB | 15.5 tok/s | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Command A (111B) | Q8_0 | 157 GB | 0.6 tok/s | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Pixtral Large 124B (2411) | Q8_0 | 173 GB | 0.6 tok/s | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Large 2 (123B) | Q8_0 | 172 GB | 0.6 tok/s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan Hy3 (295B-A21B) | Q2_K | 160 GB | 10.6 tok/s | Runs well |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3.5 Bolt E4B | Q4_K_M | 43.3 GB | 31.0 tok/s | Runs great |
| 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron Labs Dory 1.0 4B | Q4_K_M | 41.2 GB | 35.1 tok/s | Runs great |
| 60Good quality (B) · 60 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE 4.5 21B-A3B | Q4_K_M | 52.1 GB | 40.2 tok/s | Runs great |
| 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Nanbeige4.2 3B | Q4_K_M | 42.3 GB | 33.5 tok/s | Runs great |
| 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 4 E2B | Q6_K | 42.5 GB | 42.1 tok/s | Runs great |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | MiMo V2.6 Distill Qwen 9B | Q4_K_M | 44.3 GB | 14.8 tok/s | Runs well |
| 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | SenseNova Loop 9B Preview | Q4_K_M | 44.6 GB | 14.5 tok/s | Runs well |
| 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Axion 1.0 35B | Q4_K_M | 59.9 GB | 40.2 tok/s | Runs great |
| 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 4 E4B | Q4_K_M | 43.3 GB | 31.0 tok/s | Runs great |
| 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Ling 3.0 Tiny (8B-A1.3B) | Q8_0 | 46.9 GB | 45.3 tok/s | Runs great |
| 73Great quality (A) · 73 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Ornith 1.5 9B | Q4_K_M | 44.4 GB | 14.5 tok/s | Runs well |
| 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Spark-X2.6 4B | Q4_K_M | 41.1 GB | 34.0 tok/s | Runs great |
| 67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE 4.5 300B-A47B | Q3_K_M | 187 GB | 3.8 tok/s | Slow |
| 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 9B | Q4_K_M | 44.4 GB | 14.5 tok/s | Runs well |
| 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan A13B (80B-A13B) | Q4_K_M | 88.4 GB | 10.4 tok/s | Runs well |
| 59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM5 2B | Q6_K | 40.7 GB | 38.5 tok/s | Runs great |
| 71Great quality (A) · 71 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Ornith 1.0 9B | Q4_K_M | 44.4 GB | 14.5 tok/s | Runs well |
| 79Great quality (A) · 79 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE 4.5 33B | Q5_K_M | 63.4 GB | 3.4 tok/s | Slow |
| 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | TMax 9B | Q4_K_M | 44.0 GB | 15.6 tok/s | Runs well |
| 56Good quality (B) · 56 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | NorthCUA Mini | Q4_K_M | 59.9 GB | 40.2 tok/s | Runs great |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Kimi-VL A3B Thinking 2506 | Q4_K_M | 48.5 GB | 40.2 tok/s | Runs great |
| 63Good quality (B) · 63 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | dots.llm1 Instruct | Q3_K_M | 117 GB | 12.4 tok/s | Runs well |
| 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 235B-A22B Thinking 2507 | Q2_K | 134 GB | 10.1 tok/s | Runs well |
| 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 235B-A22B Instruct 2507 | Q2_K | 134 GB | 10.1 tok/s | Runs well |
| 77Great quality (A) · 77 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Seed-OSS 36B Instruct | Q5_K_M | 66.1 GB | 3.2 tok/s | Slow |
| 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | LFM3 1.7B | Q8_0 | 40.2 GB | 41.1 tok/s | Runs great |
| 54Fair quality (C) · 54 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | LFM2 24B-A2B | Q5_K_M | 55.4 GB | 42.7 tok/s | Runs great |
| 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 235B-A22B Instruct | Q2_K | 134 GB | 10.1 tok/s | Runs well |
| 76Great quality (A) · 76 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 4.3 36B | Q5_K_M | 66.1 GB | 3.2 tok/s | Slow |
| 54Fair quality (C) · 54 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Groot 19B A3B 261023 | Q4_K_M | 50.3 GB | 40.2 tok/s | Runs great |
| 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 235B-A22B | Q2_K | 134 GB | 10.1 tok/s | Runs well |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | TMax 4B | Q4_K_M | 41.1 GB | 33.2 tok/s | Runs great |
| 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | QwQ 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 32B Instruct | Q5_K_M | 64.1 GB | 3.5 tok/s | Slow |
| 75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Devstral Small 2 (24B) | Q6_K | 59.3 GB | 4.0 tok/s | Slow |
| 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 4B | Q4_K_M | 41.4 GB | 30.0 tok/s | Runs well |
| 75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.2 30B | Q6_K | 64.4 GB | 3.3 tok/s | Slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | UI-Mate 9B | Q4_K_M | 44.3 GB | 14.8 tok/s | Runs well |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | UI-Venus 2 9B | Q4_K_M | 44.3 GB | 14.8 tok/s | Runs well |
| 75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Olmo 3.1 32B Think | Q6_K | 65.4 GB | 3.0 tok/s | Slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Ling Mini 2.0 (16B-A1.4B) | Q8_0 | 55.9 GB | 42.6 tok/s | Runs great |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Agents-A1 4B | Q4_K_M | 41.3 GB | 30.8 tok/s | Runs great |
| 73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3 78B | Q6_K | 105 GB | 1.2 tok/s | Very slow |
| 73Great quality (A) · 73 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | OpenReasoning Nemotron 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 73Great quality (A) · 73 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Skywork OR1 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 73Great quality (A) · 73 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.2 8B | Q4_K_M | 45.0 GB | 15.9 tok/s | Runs well |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 8B-A1B | Q8_0 | 47.4 GB | 40.2 tok/s | Runs great |
| 73Great quality (A) · 73 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3.5 38B | Q5_K_M | 67.7 GB | 3.0 tok/s | Slow |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Mage-VL-Cowork 4.7B | Q4_K_M | 42.4 GB | 29.5 tok/s | Runs well |
| 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Baichuan M3 235B-A22B | Q2_K | 134 GB | 10.1 tok/s | Runs well |
| 73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE Deep 32B | Q6_K | 66.7 GB | 3.0 tok/s | Slow |
| 50Fair quality (C) · 50 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Coder V2 Lite (16B) | Q5_K_M | 49.7 GB | 41.2 tok/s | Runs great |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 4B Thinking 2507 | Q4_K_M | 41.9 GB | 34.7 tok/s | Runs great |
| 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Fara 1.5 4B | Q4_K_M | 41.3 GB | 30.8 tok/s | Runs great |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Snowflake Arctic Instruct | Q3_K_M | 273 GB | 10.3 tok/s | Runs well |
| 71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4 32B 0414 | Q5_K_M | 61.9 GB | 3.5 tok/s | Slow |
| 71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | GLM-Z1 32B (0414) | Q5_K_M | 61.9 GB | 3.5 tok/s | Slow |
| 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Magistral Small 1.2 (24B) | Q6_K | 59.3 GB | 4.0 tok/s | Slow |
| 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Command R+ (08-2024) 104B | Q6_K | 126 GB | 0.9 tok/s | Very slow |
| 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.1 30B | Q6_K | 64.1 GB | 3.4 tok/s | Slow |
| 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3 27B IT | Q6_K | 61.6 GB | 3.5 tok/s | Slow |
| 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE 4.0 32B | Q6_K | 65.3 GB | 3.0 tok/s | Slow |
| 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Olmo 3.1 32B Instruct | Q6_K | 65.4 GB | 3.0 tok/s | Slow |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Grok-1 (314B) | Q6_K | 298 GB | 1.1 tok/s | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Llama 4 Scout 17B-16E | Q3_K_M | 93.1 GB | 10.3 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | LFM3-VL 2B | Q6_K | 40.0 GB | 50.5 tok/s | Runs great |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 4B Instruct 2507 | Q4_K_M | 41.9 GB | 34.7 tok/s | Runs great |
| 69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 Distill Qwen 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.2 3B | Q4_K_M | 41.2 GB | 38.2 tok/s | Runs great |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Mage-VL 4.7B | Q4_K_M | 42.4 GB | 29.5 tok/s | Runs well |
| 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Devstral Small 1.1 (24B) | Q6_K | 59.0 GB | 4.1 tok/s | Slow |
| 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Magistral Small 2507 | Q6_K | 59.0 GB | 4.1 tok/s | Slow |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Kimi-VL A3B Instruct | Q4_K_M | 48.5 GB | 40.2 tok/s | Runs great |
| 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | MedGemma 27B Text IT | Q6_K | 61.2 GB | 3.6 tok/s | Slow |
| 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 0528 Qwen3 8B | Q4_K_M | 44.5 GB | 17.1 tok/s | Runs well |
| 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.0 H Small (32B-A9B) | Q4_K_M | 58.1 GB | 14.7 tok/s | Runs well |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 4B Instruct | Q4_K_M | 42.2 GB | 31.5 tok/s | Runs great |
| 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3.5 8B | Q4_K_M | 44.7 GB | 16.4 tok/s | Runs well |
| 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 8B Instruct | Q4_K_M | 44.8 GB | 15.9 tok/s | Runs well |
| 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Baichuan M2 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | OpenHands LM 32B | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 32B Instruct | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Ministral 3 14B Reasoning | Q4_K_M | 48.1 GB | 10.0 tok/s | Runs well |
| 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Magistral Small 2506 | Q6_K | 59.0 GB | 4.1 tok/s | Slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3 38B | Q5_K_M | 67.7 GB | 3.0 tok/s | Slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Small 3.2 24B | Q6_K | 59.3 GB | 4.0 tok/s | Slow |
| 67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.3 Nemotron Super 49B v1.5 | Q3_K_M | 65.4 GB | 3.6 tok/s | Slow |
| 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 Reasoning Plus (14B) | Q3_K_M | 47.1 GB | 12.3 tok/s | Runs well |
| 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Apriel 1.5 15B Thinker | Q3_K_M | 47.2 GB | 12.2 tok/s | Runs well |
| 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Terminal RSI 27B | Q6_K | 61.6 GB | 3.5 tok/s | Slow |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3.5 MoE (42B-A6.6B) | Q4_K_M | 64.9 GB | 19.8 tok/s | Runs well |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 8B | Q4_K_M | 44.5 GB | 17.1 tok/s | Runs well |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Coder V2 (236B) | Q2_K | 133 GB | 10.6 tok/s | Runs well |
| 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.2 90B Vision Instruct | Q6_K | 114 GB | 1.1 tok/s | Very slow |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3 Nano 4B | Q4_K_M | 40.8 GB | 35.2 tok/s | Runs great |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 4B | Q4_K_M | 41.9 GB | 34.7 tok/s | Runs great |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM-V 4.5 8B | Q4_K_M | 44.8 GB | 16.1 tok/s | Runs well |
| 66Good quality (B) · 66 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Hemmingway 1.5 | Q6_K | 60.9 GB | 3.6 tok/s | Slow |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.1 8B | Q4_K_M | 45.0 GB | 15.9 tok/s | Runs well |
| 65Good quality (B) · 65 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-VL 32B Instruct | Q5_K_M | 64.2 GB | 3.4 tok/s | Slow |
| 60Good quality (B) · 60 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Ministral 3 14B | Q4_K_M | 48.1 GB | 10.0 tok/s | Runs well |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron Nano 9B v2 | Q4_K_M | 43.8 GB | 15.7 tok/s | Runs well |
| 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Devstral Small 2505 | Q6_K | 59.0 GB | 4.1 tok/s | Slow |
| 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 Reasoning (14B) | Q3_K_M | 47.1 GB | 12.3 tok/s | Runs well |
| 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Small 3.1 24B | Q6_K | 59.3 GB | 4.0 tok/s | Slow |
| 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.0 H Tiny (7B-A1B) | Q8_0 | 45.7 GB | 56.0 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Moonlight 16B-A3B Instruct | Q4_K_M | 48.3 GB | 40.2 tok/s | Runs great |
| 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Unlimited OCR 2 (3B-A0.6B) | F16 | 45.5 GB | 44.1 tok/s | Runs great |
| 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 Reasoning Vision 15B | Q3_K_M | 47.4 GB | 11.9 tok/s | Runs well |
| 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Command R+ 104B (04-2024) | Q6_K | 126 GB | 0.9 tok/s | Very slow |
| 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Coder V2 Instruct (0614) | Q2_K | 133 GB | 10.6 tok/s | Runs well |
| 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V2.5 1210 | Q2_K | 133 GB | 10.6 tok/s | Runs well |
| 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | SenseNova 6.8 Code2Video | Q6_K | 61.3 GB | 3.5 tok/s | Slow |
| 64Good quality (B) · 64 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 Coder 32B Instruct | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Seed-Coder 8B Reasoning | Q4_K_M | 44.4 GB | 16.9 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | MiMo-VL 7B RL | Q4_K_M | 44.6 GB | 16.8 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.1 3B | Q4_K_M | 41.0 GB | 41.1 tok/s | Runs great |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4.1V 9B Thinking | Q4_K_M | 44.9 GB | 13.6 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | North Micro Vision | Q6_K | 41.3 GB | 39.1 tok/s | Runs great |
| 64Good quality (B) · 64 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | QwQ 32B Preview | Q5_K_M | 63.7 GB | 3.5 tok/s | Slow |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Qianfan-OCR 4B | Q4_K_M | 42.4 GB | 29.5 tok/s | Runs well |
| 57Good quality (B) · 57 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 14B | Q3_K_M | 46.9 GB | 12.2 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V2.5 | Q2_K | 133 GB | 10.6 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1R 7B | Q4_K_M | 43.3 GB | 18.4 tok/s | Runs well |
| 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Small 3 24B (2501) | Q6_K | 59.0 GB | 4.1 tok/s | Slow |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 2.6B | Q5_K_M | 40.3 GB | 42.7 tok/s | Runs great |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | MiMo 7B RL | Q4_K_M | 44.3 GB | 17.8 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE Deep 7.8B | Q4_K_M | 44.1 GB | 17.9 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | GLM-Z1 9B (0414) | Q4_K_M | 44.4 GB | 14.9 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3 8B | Q4_K_M | 43.6 GB | 17.6 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 Mini Reasoning (3.8B) | Q4_K_M | 41.7 GB | 36.4 tok/s | Runs great |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3 12B IT | Q4_K_M | 46.4 GB | 11.4 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Seed-Coder 8B Instruct | Q4_K_M | 44.4 GB | 16.9 tok/s | Runs well |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM-OCR 1.0 1.3B | Q8_0 | 39.8 GB | 54.7 tok/s | Runs great |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | PaddleOCR-VL 1.6 0.9B | Q8_0 | 39.5 GB | 74.1 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3n E4B | Q4_K_M | 43.2 GB | 34.9 tok/s | Runs great |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 Coder 7B Instruct | Q4_K_M | 43.4 GB | 18.3 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5-VL 3B | Q4_K_M | 40.3 GB | 44.8 tok/s | Runs great |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Jamba Reasoning 3B | Q4_K_M | 40.2 GB | 43.6 tok/s | Runs great |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | LightOnOCR-3 4B | Q4_K_M | 41.3 GB | 30.8 tok/s | Runs great |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Olmo 3 7B Think | Q4_K_M | 44.2 GB | 19.1 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | AceReason Nemotron 14B | Q3_K_M | 47.1 GB | 12.2 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | OpenReasoning Nemotron 14B | Q3_K_M | 47.1 GB | 12.2 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 Coder 14B Instruct | Q3_K_M | 47.1 GB | 12.2 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 4 14B | Q3_K_M | 46.9 GB | 12.2 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Ministral 3 8B | Q4_K_M | 44.9 GB | 15.7 tok/s | Runs well |
| 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Apriel Nemotron 15B Thinker | Q3_K_M | 47.3 GB | 12.1 tok/s | Runs well |
| 62Good quality (B) · 62 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1 34B Instruct | Q5_K_M | 63.4 GB | 3.4 tok/s | Slow |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | OpenReasoning Nemotron 7B | Q4_K_M | 43.4 GB | 18.3 tok/s | Runs well |
| 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Dolphin Mistral 24B Venice Edition | Q6_K | 59.3 GB | 4.0 tok/s | Slow |
| 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Voxtral Small 24B | Q6_K | 59.5 GB | 4.0 tok/s | Slow |
| 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Unlimited OCR (3B-A0.6B) | F16 | 45.5 GB | 44.1 tok/s | Runs great |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 Mini (3.8B) | Q4_K_M | 41.7 GB | 36.4 tok/s | Runs great |
| 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 2 27B IT | Q6_K | 62.3 GB | 3.6 tok/s | Slow |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 Nemotron Nano 8B v1 | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Ministral 3 3B | Q4_K_M | 41.5 GB | 36.3 tok/s | Runs great |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM4 8B | Q4_K_M | 43.5 GB | 17.1 tok/s | Runs well |
| 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | TeleOCR | Q8_0 | 40.7 GB | 50.1 tok/s | Runs great |
| 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Command R (08-2024) 32B | Q6_K | 66.1 GB | 3.0 tok/s | Slow |
| 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM-V 4.7 1B | Q8_0 | 40.0 GB | 45.9 tok/s | Runs great |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM3 3B | Q4_K_M | 40.8 GB | 45.3 tok/s | Runs great |
| 40Fair quality (C) · 40 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Jamba2 3B | Q5_K_M | 40.4 GB | 38.1 tok/s | Runs great |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V2 Lite Chat | Q5_K_M | 49.7 GB | 41.2 tok/s | Runs great |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 (14B) | Q3_K_M | 47.2 GB | 12.3 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | DeepCoder 14B Preview | Q3_K_M | 47.1 GB | 12.2 tok/s | Runs well |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 2B | Q6_K | 40.2 GB | 42.7 tok/s | Runs great |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | TMax 2B | Q6_K | 39.9 GB | 51.5 tok/s | Runs great |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3 4B IT | Q4_K_M | 41.1 GB | 32.5 tok/s | Runs great |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Chandra OCR 2 | Q4_K_M | 41.8 GB | 26.4 tok/s | Runs well |
| 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | InternVL3 14B | Q3_K_M | 47.3 GB | 11.9 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan 7B Instruct | Q4_K_M | 43.9 GB | 18.6 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 7B Instruct | Q4_K_M | 43.4 GB | 18.3 tok/s | Runs well |
| 59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Dolphin 3.0 R1 Mistral 24B | Q6_K | 59.0 GB | 4.1 tok/s | Slow |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | LFM2 8B-A1B | Q8_0 | 47.2 GB | 40.2 tok/s | Runs great |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | MedGemma 4B IT | Q4_K_M | 41.1 GB | 32.5 tok/s | Runs great |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Foundation-Sec 8B Reasoning | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Mixtral 8x22B Instruct | Q4_K_M | 126 GB | 3.5 tok/s | Slow |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-VL 7B Instruct | Q4_K_M | 43.8 GB | 16.8 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.0 Micro (3B) | Q4_K_M | 41.0 GB | 41.1 tok/s | Runs great |
| 59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Aya Expanse 32B | Q6_K | 66.1 GB | 3.0 tok/s | Slow |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 Nemotron Nano 4B v1.1 | Q4_K_M | 42.1 GB | 31.0 tok/s | Runs great |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | WeVisOne-VL 4B | Q4_K_M | 42.4 GB | 28.9 tok/s | Runs well |
| 59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | OLMo 2 32B Instruct | Q6_K | 65.8 GB | 3.0 tok/s | Slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 14B Instruct | Q3_K_M | 47.1 GB | 12.3 tok/s | Runs well |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 Distill Qwen 14B | Q3_K_M | 47.1 GB | 12.2 tok/s | Runs well |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | JanusCoder 14B | Q3_K_M | 46.9 GB | 12.2 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Olmo 3 7B Instruct | Q4_K_M | 44.2 GB | 19.1 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM-o 2.6 8B | Q4_K_M | 44.0 GB | 16.1 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V2 Chat | Q2_K | 133 GB | 10.6 tok/s | Runs well |
| 58Good quality (B) · 58 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Aya Vision 32B | Q5_K_M | 63.2 GB | 3.5 tok/s | Slow |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1 7B Instruct | Q4_K_M | 43.3 GB | 18.4 tok/s | Runs well |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan 4B Instruct | Q4_K_M | 42.1 GB | 33.1 tok/s | Runs great |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 7B Instruct 1M | Q4_K_M | 43.4 GB | 18.3 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3.5 Mini Instruct | Q4_K_M | 43.8 GB | 36.6 tok/s | Runs great |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4 9B 0414 | Q4_K_M | 44.4 GB | 14.9 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | NorthCUA Nano | Q4_K_M | 44.3 GB | 14.8 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | GLiDE 9B | Q4_K_M | 43.4 GB | 17.6 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 Distill Llama 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Foundation-Sec 1.1 8B Instruct | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | InternVL2.5 8B | Q4_K_M | 44.3 GB | 17.3 tok/s | Runs well |
| 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 14B Instruct 1M | Q3_K_M | 47.1 GB | 12.2 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Fara 7B | Q4_K_M | 43.8 GB | 16.8 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Nous Hermes 2 Mixtral 8x7B DPO | Q4_K_M | 67.8 GB | 10.4 tok/s | Runs well |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.0 H Micro | Q4_K_M | 40.3 GB | 43.8 tok/s | Runs great |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 3 10B Instruct | Q4_K_M | 45.9 GB | 13.5 tok/s | Runs well |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM5 1B | Q8_0 | 39.6 GB | 65.8 tok/s | Runs great |
| 35Basic quality (D) · 35 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3n E2B | Q6_K | 42.9 GB | 48.5 tok/s | Runs great |
| 35Basic quality (D) · 35 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3-VL 2B Instruct | Q6_K | 41.0 GB | 45.5 tok/s | Runs great |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | InternLM3 8B Instruct | Q4_K_M | 44.1 GB | 15.9 tok/s | Runs well |
| 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Kimi Dev 72B | Q2_K | 70.1 GB | 3.2 tok/s | Slow |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 2 9B IT | Q4_K_M | 45.5 GB | 15.1 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Phi-4 Multimodal Instruct | Q4_K_M | 42.8 GB | 25.1 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Jamba 1.5 Mini | Q4_K_M | 69.9 GB | 11.2 tok/s | Runs well |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 3 (Llama 3.1 8B) | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 8B Instruct | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Nemo 12B Instruct | Q4_K_M | 47.1 GB | 11.4 tok/s | Runs well |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM-V 2.6 8B | Q4_K_M | 43.7 GB | 17.2 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1 3B Instruct | Q4_K_M | 40.5 GB | 44.3 tok/s | Runs great |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 Coder 3B Instruct | Q4_K_M | 40.5 GB | 45.2 tok/s | Runs great |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Mixtral 8x7B Instruct | Q4_K_M | 67.8 GB | 10.4 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | AFM 4.5B | Q4_K_M | 41.7 GB | 30.2 tok/s | Runs great |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3.5 Vision Instruct | Q4_K_M | 44.0 GB | 33.7 tok/s | Runs great |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | MiniCPM3 4B | Q4_K_M | 41.1 GB | 34.3 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 3 7B Instruct | Q4_K_M | 43.8 GB | 18.7 tok/s | Runs well |
| 33Basic quality (D) · 33 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.2 3B Instruct | Q4_K_M | 41.2 GB | 43.5 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 Distill Qwen 7B | Q4_K_M | 43.4 GB | 18.3 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3 Mini 128K Instruct | Q4_K_M | 43.8 GB | 36.6 tok/s | Runs great |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Qwen1.5 110B Chat | Q6_K | 132 GB | 0.9 tok/s | Very slow |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE 3.5 7.8B | Q4_K_M | 44.1 GB | 17.9 tok/s | Runs well |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4 9B Chat (2024) | Q4_K_M | 44.4 GB | 14.9 tok/s | Runs well |
| 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | GLM-4 9B Chat 1M | Q4_K_M | 44.7 GB | 14.7 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | LFM2 2.6B | Q5_K_M | 40.2 GB | 44.9 tok/s | Runs great |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3 Mini 4K Instruct | Q4_K_M | 42.2 GB | 36.6 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Ministral 8B (2410) | Q4_K_M | 44.4 GB | 17.4 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | DeepHermes 3 Llama 3 8B Preview | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Dolphin 3.0 Llama 3.1 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Tülu 3 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.2 8B Instruct | Q4_K_M | 44.6 GB | 17.1 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.3 8B | Q4_K_M | 44.6 GB | 17.1 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Pixtral 12B (2409) | Q4_K_M | 47.2 GB | 11.3 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 3B Instruct | Q4_K_M | 40.5 GB | 45.2 tok/s | Runs great |
| 52Fair quality (C) · 52 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Command R 35B (v01) | Q5_K_M | 73.9 GB | 3.3 tok/s | Slow |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 1.2B Instruct | Q8_0 | 39.6 GB | 60.8 tok/s | Runs great |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3 Small 8K Instruct | Q4_K_M | 43.9 GB | 18.9 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Apertus v1.5 8B | Q4_K_M | 44.8 GB | 15.7 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2 7B Instruct | Q4_K_M | 43.4 GB | 18.3 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-VL 3B Instruct | Q4_K_M | 40.9 GB | 37.2 tok/s | Runs great |
| 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Reka Flash 3.1 (21B) | Q2_K | 47.8 GB | 11.0 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | CodeGeeX4 All 9B | Q4_K_M | 44.4 GB | 14.9 tok/s | Runs well |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 4 70B | Q2_K | 69.2 GB | 3.3 tok/s | Slow |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.3 70B Instruct | Q2_K | 69.2 GB | 3.2 tok/s | Slow |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Dolphin 2.9.4 Llama 3.1 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 2 Pro Llama 3 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | OpenChat 3.6 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | InternVL2 8B | Q4_K_M | 44.3 GB | 17.3 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Command R7B | Q4_K_M | 43.6 GB | 17.4 tok/s | Runs well |
| 35Basic quality (D) · 35 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | S5 nano | Q4_K_M | 41.4 GB | 30.0 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.1 8B Instruct | Q4_K_M | 44.6 GB | 17.1 tok/s | Runs well |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 3 Llama 3.2 3B | Q4_K_M | 41.2 GB | 43.5 tok/s | Runs great |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2-VL 7B Instruct | Q4_K_M | 43.8 GB | 16.8 tok/s | Runs well |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 72B Instruct | Q2_K | 70.0 GB | 3.2 tok/s | Slow |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-VL 72B Instruct | Q2_K | 70.3 GB | 3.1 tok/s | Slow |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2 57B-A14B Instruct | Q3_K_M | 66.9 GB | 12.4 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Aya Vision 8B | Q4_K_M | 43.9 GB | 16.2 tok/s | Runs well |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE 4.0 1.2B | Q8_0 | 40.1 GB | 55.5 tok/s | Runs great |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3 Vision 128K Instruct | Q4_K_M | 44.0 GB | 33.7 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-Omni 7B | Q4_K_M | 45.3 GB | 13.0 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Voxtral Mini 3B | Q4_K_M | 42.1 GB | 29.8 tok/s | Runs well |
| 29Basic quality (D) · 29 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE 3.5 2.4B | Q6_K | 40.9 GB | 40.2 tok/s | Runs great |
| 29Basic quality (D) · 29 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 1.7B | Q6_K | 40.9 GB | 47.7 tok/s | Runs great |
| 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 Distill Llama 70B | Q2_K | 69.2 GB | 3.3 tok/s | Slow |
| 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Tülu 3 70B | Q2_K | 69.2 GB | 3.3 tok/s | Slow |
| 29Basic quality (D) · 29 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.2 2B Instruct | Q5_K_M | 40.8 GB | 45.6 tok/s | Runs great |
| 29Basic quality (D) · 29 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.3 2B | Q5_K_M | 40.8 GB | 45.6 tok/s | Runs great |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Prover V2 7B | Q4_K_M | 46.5 GB | 20.2 tok/s | Runs well |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3 8B Instruct | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Command R7B Arabic | Q4_K_M | 43.6 GB | 17.4 tok/s | Runs well |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 3 3B Instruct | Q4_K_M | 41.0 GB | 43.2 tok/s | Runs great |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Mistral-NeMo-Minitron 8B Instruct | Q4_K_M | 44.8 GB | 16.6 tok/s | Runs well |
| 48Fair quality (C) · 48 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Aya 23 35B | Q5_K_M | 73.9 GB | 3.3 tok/s | Slow |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 70B Instruct | Q2_K | 69.2 GB | 3.2 tok/s | Slow |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.0 H 1B | Q8_0 | 39.9 GB | 48.7 tok/s | Runs great |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Codestral 22B | Q2_K | 49.1 GB | 10.3 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3 Medium 128K Instruct | Q4_K_M | 48.5 GB | 10.0 tok/s | Runs well |
| 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Phi-3 Medium (14B) | Q4_K_M | 47.6 GB | 10.0 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | InternLM2.5 7B Chat | Q4_K_M | 44.1 GB | 18.0 tok/s | Runs well |
| 48Fair quality (C) · 48 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Yi 1.5 34B Chat | Q5_K_M | 63.7 GB | 3.4 tok/s | Slow |
| 27Basic quality (D) · 27 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.1 2B Instruct | Q5_K_M | 40.8 GB | 45.6 tok/s | Runs great |
| 27Basic quality (D) · 27 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 2 2B | Q5_K_M | 40.6 GB | 44.2 tok/s | Runs great |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Aya Expanse 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Apertus 8B Instruct | Q4_K_M | 44.3 GB | 17.3 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 3 (Llama 3.1 70B) | Q2_K | 69.2 GB | 3.3 tok/s | Slow |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.0 8B Instruct | Q4_K_M | 43.9 GB | 17.1 tok/s | Runs well |
| 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | QVQ 72B Preview | Q2_K | 70.3 GB | 3.1 tok/s | Slow |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | WeVisOne VL 2B | Q6_K | 41.2 GB | 39.7 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | OLMoE 1B-7B Instruct | Q8_0 | 46.2 GB | 45.3 tok/s | Runs great |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Solar Pro Preview 22B | Q2_K | 48.5 GB | 10.4 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | OLMo 2 13B Instruct | Q4_K_M | 50.0 GB | 10.2 tok/s | Runs well |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Mistral 7B Instruct v0.3 | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.2 11B Vision Instruct | Q4_K_M | 46.1 GB | 13.1 tok/s | Runs well |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Falcon3 Mamba 7B Instruct | Q4_K_M | 42.8 GB | 19.2 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Yi Coder 9B Chat | Q4_K_M | 44.5 GB | 15.8 tok/s | Runs well |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Zamba2 7B Instruct | Q4_K_M | 43.5 GB | 19.0 tok/s | Runs well |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | DeepScaleR 1.5B Preview | Q8_0 | 40.4 GB | 39.9 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 Coder 1.5B Instruct | Q8_0 | 40.2 GB | 46.2 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1 1.5B Instruct | Q8_0 | 40.1 GB | 45.9 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | LFM2 1.2B | Q8_0 | 39.6 GB | 60.8 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.5 0.8B | Q8_0 | 39.3 GB | 81.7 tok/s | Runs great |
| 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek MoE 16B Chat | Q4_K_M | 49.2 GB | 42.7 tok/s | Runs great |
| 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 Nemotron 70B Instruct | Q2_K | 69.2 GB | 3.3 tok/s | Slow |
| 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Apertus v1.5 70B | Q2_K | 69.8 GB | 3.2 tok/s | Slow |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | EXAONE 3.0 7.8B Instruct | Q4_K_M | 43.6 GB | 17.9 tok/s | Runs well |
| 25Basic quality (D) · 25 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.0 2B Instruct | Q5_K_M | 40.5 GB | 43.9 tok/s | Runs great |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Dolphin 2.9 Llama 3 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | LLM4Decompile 22B V2 | Q2_K | 49.1 GB | 10.3 tok/s | Runs well |
| 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Granite Vision 3.2 2B | Q4_K_M | 40.8 GB | 46.9 tok/s | Runs great |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | OLMo 2 7B Instruct | Q4_K_M | 44.9 GB | 19.1 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Hermes 2 Pro Mistral 7B | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Starling LM 7B Beta | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Nous Hermes 2 SOLAR 10.7B | Q4_K_M | 45.6 GB | 13.0 tok/s | Runs well |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | SOLAR 10.7B Instruct | Q4_K_M | 45.6 GB | 13.0 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Codestral Mamba 7B | Q4_K_M | 42.8 GB | 19.2 tok/s | Runs well |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | LLM4Decompile 9B V2 | Q4_K_M | 44.5 GB | 15.8 tok/s | Runs well |
| 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2 72B Instruct | Q2_K | 70.1 GB | 3.2 tok/s | Slow |
| 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2-VL 72B Instruct | Q2_K | 70.3 GB | 3.1 tok/s | Slow |
| 24Basic quality (D) · 24 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 1.5B Instruct | Q8_0 | 40.2 GB | 46.2 tok/s | Runs great |
| 24Basic quality (D) · 24 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | LFM2-VL 1.6B | Q8_0 | 40.1 GB | 45.0 tok/s | Runs great |
| 23Basic quality (D) · 23 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | SmolVLM2 2.2B Instruct | Q6_K | 41.7 GB | 43.1 tok/s | Runs great |
| 23Basic quality (D) · 23 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2-VL 2B Instruct | Q6_K | 40.3 GB | 43.8 tok/s | Runs great |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-Math 7B Instruct | Q4_K_M | 43.2 GB | 18.3 tok/s | Runs well |
| 22Basic quality (D) · 22 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Qwen1.5 MoE A2.7B Chat | Q4_K_M | 48.6 GB | 44.1 tok/s | Runs great |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Coder 6.7B Instruct | Q4_K_M | 46.7 GB | 20.7 tok/s | Runs well |
| 33Basic quality (D) · 33 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | OpenChat 3.5 (0106) | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Mistral 7B Instruct v0.2 | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Nous Hermes 2 Mistral 7B DPO | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | OpenChat 3.5 (1210) | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Mathstral 7B | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 37Basic quality (D) · 37 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Mistral Small 22B (2409) | Q2_K | 49.1 GB | 10.3 tok/s | Runs well |
| 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Yi 1.5 9B Chat | Q4_K_M | 44.1 GB | 15.8 tok/s | Runs well |
| 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-Omni 3B | Q3_K_M | 41.3 GB | 32.6 tok/s | Runs great |
| 22Basic quality (D) · 22 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek R1 Distill Qwen 1.5B | Q8_0 | 40.4 GB | 39.9 tok/s | Runs great |
| 22Basic quality (D) · 22 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.2 1B Instruct | Q8_0 | 39.9 GB | 57.3 tok/s | Runs great |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | SmolVLM 2.2B Instruct | Q6_K | 41.7 GB | 43.1 tok/s | Runs great |
| 22Basic quality (D) · 22 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan 1.8B Instruct | Q8_0 | 40.7 GB | 39.7 tok/s | Runs great |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | LLM4Decompile 6.7B V2 | Q4_K_M | 46.7 GB | 20.7 tok/s | Runs well |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Qwen1.5 32B Chat | Q4_K_M | 60.3 GB | 4.3 tok/s | Slow |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Coder 33B Instruct | Q4_K_M | 60.7 GB | 4.2 tok/s | Slow |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Aya 23 8B | Q4_K_M | 44.3 GB | 17.4 tok/s | Runs well |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Zamba2 2.7B Instruct | Q5_K_M | 40.5 GB | 43.4 tok/s | Runs great |
| 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Apertus 70B Instruct | Q2_K | 69.2 GB | 3.2 tok/s | Slow |
| 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | LLaVA 1.6 34B | Q4_K_M | 60.5 GB | 4.0 tok/s | Slow |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | CodeGemma 7B | Q4_K_M | 47.3 GB | 16.4 tok/s | Runs well |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | d1 2B | Q6_K | 40.0 GB | 50.5 tok/s | Runs great |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | OpenHermes 2.5 Mistral 7B | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Starling LM 7B Alpha | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | InternLM2.5 20B Chat | Q2_K | 47.8 GB | 11.5 tok/s | Runs well |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.1 3B-A800M Instruct | Q8_0 | 42.3 GB | 66.4 tok/s | Runs great |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM2 1.7B Instruct | Q8_0 | 41.7 GB | 41.6 tok/s | Runs great |
| 33Basic quality (D) · 33 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 180B Chat | Q6_K | 186 GB | 0.5 tok/s | Very slow |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-Math 1.5B Instruct | Q8_0 | 40.0 GB | 46.2 tok/s | Runs great |
| 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3 70B Instruct | Q2_K | 69.2 GB | 3.3 tok/s | Slow |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3 1B | Q8_0 | 39.4 GB | 71.1 tok/s | Runs great |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | InternLM2 Math Plus 7B | Q4_K_M | 44.1 GB | 18.0 tok/s | Runs well |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | SQLCoder 7B v2 | Q4_K_M | 46.7 GB | 20.7 tok/s | Runs well |
| 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5-Math 72B Instruct | Q2_K | 68.7 GB | 3.2 tok/s | Slow |
| 19Basic quality (D) · 19 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Stable Code Instruct 3B | Q5_K_M | 43.0 GB | 41.2 tok/s | Runs great |
| 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | RecurrentGemma 9B IT | Q4_K_M | 44.3 GB | 14.5 tok/s | Runs well |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Granite Code 8B Instruct (128K) | Q4_K_M | 44.4 GB | 17.3 tok/s | Runs well |
| 19Basic quality (D) · 19 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Granite Code 3B Instruct (128K) | Q4_K_M | 43.1 GB | 40.1 tok/s | Runs great |
| 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Yi 1.5 6B Chat | Q3_K_M | 41.5 GB | 29.8 tok/s | Runs well |
| 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | StarCoder2 15B Instruct | Q3_K_M | 46.8 GB | 11.3 tok/s | Runs well |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Mistral 7B Instruct v0.1 | Q4_K_M | 43.8 GB | 19.3 tok/s | Runs well |
| 18Basic quality (D) · 18 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron Mini 4B Instruct | Q3_K_M | 40.9 GB | 43.1 tok/s | Runs great |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Falcon Mamba 7B Instruct | Q4_K_M | 42.8 GB | 19.2 tok/s | Runs well |
| 18Basic quality (D) · 18 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Granite 3.0 3B-A800M Instruct | Q8_0 | 42.1 GB | 66.4 tok/s | Runs great |
| 18Basic quality (D) · 18 of 100: how capable the model is at F16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | LFM2 700M | F16 | 39.9 GB | 44.3 tok/s | Runs great |
| 17Basic quality (D) · 17 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | InternLM2.5 1.8B Chat | Q6_K | 40.6 GB | 51.3 tok/s | Runs great |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Yi 34B Chat | Q4_K_M | 60.3 GB | 4.1 tok/s | Slow |
| 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Granite Code 34B Instruct | Q4_K_M | 59.2 GB | 4.1 tok/s | Slow |
| 17Basic quality (D) · 17 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | StarCoder2 3B | Q4_K_M | 40.4 GB | 46.1 tok/s | Runs great |
| 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Neural Chat 7B v3.3 | Q3_K_M | 42.9 GB | 24.9 tok/s | Runs well |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | GLiDE 2B | Q6_K | 39.9 GB | 51.5 tok/s | Runs great |
| 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | LLaVA 1.6 Mistral 7B | Q3_K_M | 43.1 GB | 23.9 tok/s | Runs well |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Yi Coder 1.5B Chat | Q8_0 | 41.5 GB | 48.0 tok/s | Runs great |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 3 1B Instruct | Q8_0 | 40.7 GB | 42.6 tok/s | Runs great |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Spark X2.6 1.7B | Q8_0 | 40.2 GB | 41.6 tok/s | Runs great |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2 1.5B Instruct | Q8_0 | 40.2 GB | 46.2 tok/s | Runs great |
| 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | Code Llama 34B Instruct | Q4_K_M | 60.5 GB | 4.1 tok/s | Slow |
| 15Basic quality (D) · 15 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | StableLM Zephyr 3B | Q5_K_M | 41.6 GB | 41.2 tok/s | Runs great |
| 15Basic quality (D) · 15 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Phi-2 2.7B | Q5_K_M | 40.9 GB | 41.5 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 1.1 7B IT | Q4_K_M | 47.3 GB | 16.4 tok/s | Runs well |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Qwen1.5 14B Chat | Q3_K_M | 51.9 GB | 12.7 tok/s | Runs well |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | WizardMath 7B v1.1 | Q3_K_M | 42.9 GB | 24.9 tok/s | Runs well |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Zephyr 7B Beta | Q3_K_M | 42.9 GB | 24.9 tok/s | Runs well |
| 15Basic quality (D) · 15 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | surya ocr 2.5 | F16 | 39.8 GB | 47.5 tok/s | Runs great |
| 29Basic quality (D) · 29 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | InternLM2 Chat 20B | Q2_K | 47.8 GB | 11.5 tok/s | Runs well |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Code Llama 7B Instruct | Q2_K | 45.3 GB | 34.0 tok/s | Runs great |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Qwen1.5 7B Chat | Q3_K_M | 46.4 GB | 23.4 tok/s | Runs well |
| 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 0.6B | F16 | 40.7 GB | 43.7 tok/s | Runs great |
| 14Basic quality (D) · 14 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM 1.7B Instruct | Q8_0 | 40.5 GB | 41.6 tok/s | Runs great |
| 14Basic quality (D) · 14 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | OLMo 2 1B Instruct | Q8_0 | 40.4 GB | 48.0 tok/s | Runs great |
| 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1 0.5B Instruct | F16 | 39.5 GB | 63.0 tok/s | Runs great |
| 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 350M | F16 | 39.1 GB | 93.6 tok/s | Runs great |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | Code Llama 13B Instruct | Q4_K_M | 52.9 GB | 10.7 tok/s | Runs well |
| 13Basic quality (D) · 13 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | CodeGemma 2B | Q5_K_M | 40.2 GB | 46.0 tok/s | Runs great |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | StarCoder2 7B | Q2_K | 41.7 GB | 32.0 tok/s | Runs great |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | InternLM2 Math Plus 20B | Q2_K | 47.8 GB | 11.5 tok/s | Runs well |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 7B IT | Q3_K_M | 46.2 GB | 21.1 tok/s | Runs well |
| 14Basic quality (D) · 14 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Yi 6B Chat | Q2_K | 41.0 GB | 37.9 tok/s | Runs great |
| 15Basic quality (D) · 15 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Orca 2 7B | Q2_K | 43.1 GB | 34.0 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | CogVLM2 Llama 3 Chat 19B | Q2_K | 50.4 GB | 11.8 tok/s | Runs well |
| 12Basic quality (D) · 12 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek Coder 1.3B Instruct | Q8_0 | 41.3 GB | 52.7 tok/s | Runs great |
| 12Basic quality (D) · 12 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | LLM4Decompile 1.3B V2 | Q8_0 | 41.3 GB | 52.7 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | StarCoder2 15B (base) | Q3_K_M | 46.8 GB | 11.3 tok/s | Runs well |
| 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | LFM2-VL 450M | F16 | 39.3 GB | 72.8 tok/s | Runs great |
| 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | LFM2 350M | F16 | 39.1 GB | 93.6 tok/s | Runs great |
| 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Granite 4.0 H 350M | F16 | 39.0 GB | 96.4 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | StableLM 2 12B Chat | Q4_K_M | 46.5 GB | 11.5 tok/s | Runs well |
| 14Basic quality (D) · 14 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Llama 2 7B Chat | Q2_K | 43.1 GB | 34.0 tok/s | Runs great |
| 11Basic quality (D) · 11 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 2B IT | Q5_K_M | 40.2 GB | 46.0 tok/s | Runs great |
| 26Basic quality (D) · 26 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Orca 2 13B | Q4_K_M | 49.6 GB | 10.7 tok/s | Runs well |
| 11Basic quality (D) · 11 of 100: how capable the model is at F16, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | GLiDE 0.8B | F16 | 39.9 GB | 43.7 tok/s | Runs great |
| 25Basic quality (D) · 25 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | Granite Code 20B Instruct | Q2_K | 46.5 GB | 11.4 tok/s | Runs well |
| 14Basic quality (D) · 14 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | Baichuan 2 7B Chat | Q2_K | 43.4 GB | 30.5 tok/s | Runs great |
| 10Basic quality (D) · 10 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | StableLM 2 1.6B Chat | Q8_0 | 40.8 GB | 43.4 tok/s | Runs great |
| 10Basic quality (D) · 10 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | StableLM 2 Zephyr 1.6B | Q8_0 | 40.8 GB | 43.4 tok/s | Runs great |
| 30Basic quality (D) · 30 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | Code Llama 70B Instruct | Q2_K | 67.2 GB | 3.3 tok/s | Slow |
| 12Basic quality (D) · 12 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek LLM 7B Chat | Q2_K | 43.1 GB | 33.2 tok/s | Runs great |
| 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | SmolVLM 500M Instruct | F16 | 39.6 GB | 64.3 tok/s | Runs great |
| 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 0.5B Instruct | F16 | 39.4 GB | 66.9 tok/s | Runs great |
| 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 230M | F16 | 38.8 GB | 142 tok/s | Runs great |
| 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | Llama 2 13B Chat | Q4_K_M | 49.6 GB | 10.7 tok/s | Runs well |
| 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | Baichuan 2 13B Chat | Q4_K_M | 50.1 GB | 10.0 tok/s | Runs well |
| 11Basic quality (D) · 11 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | LLaVA 1.5 7B | Q2_K | 43.3 GB | 32.5 tok/s | Runs great |
| 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan 0.5B Instruct | F16 | 40.2 GB | 60.7 tok/s | Runs great |
| 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2.5 Coder 0.5B Instruct | F16 | 39.4 GB | 66.9 tok/s | Runs great |
| 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM2 360M Instruct | F16 | 39.3 GB | 91.0 tok/s | Runs great |
| 9Basic quality (D) · 9 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | Vicuna 7B v1.5 | Q2_K | 43.1 GB | 34.0 tok/s | Runs great |
| 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek LLM 67B Chat | Q2_K | 66.8 GB | 3.4 tok/s | Slow |
| 9Basic quality (D) · 9 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | OLMo 7B Instruct | Q2_K | 42.1 GB | 33.3 tok/s | Runs great |
| 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | StarCoder 15B | Q3_K_M | 46.2 GB | 11.4 tok/s | Runs well |
| 6Basic quality (D) · 6 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | TinyLlama 1.1B Chat | Q8_0 | 39.5 GB | 64.6 tok/s | Runs great |
| 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | Qwen2 0.5B Instruct | F16 | 39.4 GB | 66.9 tok/s | Runs great |
| 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE 4.5 0.3B | F16 | 39.2 GB | 91.0 tok/s | Runs great |
| 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | SmolVLM 256M Instruct | F16 | 39.0 GB | 126 tok/s | Runs great |
| 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 3 270M | F16 | 38.9 GB | 121 tok/s | Runs great |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | WizardLM 13B v1.2 | Q4_K_M | 49.6 GB | 10.7 tok/s | Runs well |
| 20Basic quality (D) · 20 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | LLaVA 1.5 13B | Q4_K_M | 49.8 GB | 10.5 tok/s | Runs well |
| 5Basic quality (D) · 5 of 100: how capable the model is at F16, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM 360M Instruct | F16 | 39.1 GB | 91.0 tok/s | Runs great |
| 8Basic quality (D) · 8 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | InternLM 7B | Q2_K | 42.3 GB | 31.3 tok/s | Runs great |
| 25Basic quality (D) · 25 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 43/100 (quality, fit, speed and memory use) sets the ranking. | Llama 2 70B Chat | Q2_K | 67.2 GB | 3.3 tok/s | Slow |
| 16Basic quality (D) · 16 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 43/100 (quality, fit, speed and memory use) sets the ranking. | Vicuna 13B v1.5 | Q3_K_M | 48.0 GB | 13.9 tok/s | Runs well |
| 4Basic quality (D) · 4 of 100: how capable the model is at F16, whatever it runs on. Match score 43/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM2 135M Instruct | F16 | 38.7 GB | 252 tok/s | Runs great |
| 3Basic quality (D) · 3 of 100: how capable the model is at F16, whatever it runs on. Match score 43/100 (quality, fit, speed and memory use) sets the ranking. | Falcon-H1 Tiny 90M Instruct | F16 | 38.6 GB | 364 tok/s | Runs great |
| 2Basic quality (D) · 2 of 100: how capable the model is at F16, whatever it runs on. Match score 42/100 (quality, fit, speed and memory use) sets the ranking. | SmolLM 135M Instruct | F16 | 38.6 GB | 252 tok/s | Runs great |
| 4Basic quality (D) · 4 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 42/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 7B Instruct | Q2_K | 41.2 GB | 31.8 tok/s | Runs great |
| 15Basic quality (D) · 15 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 39/100 (quality, fit, speed and memory use) sets the ranking. | Falcon 40B Instruct | Q4_K_M | 64.0 GB | 3.3 tok/s | Slow |
| 69Good quality (B) · 69 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K3.1 Pro (2.8T) | Q2_K | 1,151 GB | — | Won't run |
| 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3.8 Max (2.4T-A95B) | Q2_K | 1,018 GB | — | Won't run |
| 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | Kimi K3 (2.8T) | Q2_K | 1,150 GB | — | Won't run |
| 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V4 Pro (1.6T-A49B) | Q2_K | 699 GB | — | Won't run |
| 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | DeepSeek V4 Pro (April 2026) | Q2_K | 678 GB | — | Won't run |
Image models on the AMD EPYC 7002 / 7003 series NAS (512 GB)
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | HunyuanImage 3.0 (80B-A13B) | FP16 | 170 GB | 41.1 min | Very slow |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Cosmos3-Super Text2Image (64B) | FP16 | 133 GB | 27.4 min | Very slow |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.2 [dev] | FP16 | 116 GB | 24.7 min | Very slow |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Krea 2 Turbo | FP16 | 37.4 GB | 2.9 min | Very slow |
| 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Ideogram 4 | FP16 | 57.3 GB | 17.1 min | Very slow |
| 93Excellent quality (A+) · 93 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Qwen-Image 2.1 | FP16 | 57.0 GB | 41.1 min | Very slow |
| 91Excellent quality (A+) · 91 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | FLUX 3 Image Klein | FP16 | 22.4 GB | 3.9 min | Very slow |
| 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.2 [klein] 9B | FP16 | 37.1 GB | 74 s | Slow |
| 93Excellent quality (A+) · 93 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Fibo 2.0 Dev | FP16 | 27.7 GB | 15.4 min | Very slow |
| 89Excellent quality (A+) · 89 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Qwen-Image 2.1 Turbo | FP16 | 46.7 GB | 7.9 min | Very slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Qwen-Image 2512 | FP16 | 63.8 GB | 61.6 min | Very slow |
| 85Excellent quality (A+) · 85 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE-Image Turbo | FP16 | 26.4 GB | 2.2 min | Slow |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Ming-Image 0.1 Design | FP16 | 63.9 GB | 11.2 min | Very slow |
| 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Z-Image Turbo | FP16 | 22.7 GB | 1.7 min | Slow |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | Qwen-Image | FP16 | 63.8 GB | 61.6 min | Very slow |
| 87Excellent quality (A+) · 87 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | FireRed-Image-Edit 1.1 | FP16 | 61.2 GB | 54.8 min | Very slow |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE-Image | FP16 | 34.0 GB | 15.1 min | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP8, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.2 [klein] 4B | FP8 | 10.2 GB | 37 s | Runs well |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Qwen-Image-Edit 2511 | FP16 | 61.2 GB | 54.8 min | Very slow |
| 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | HiDream-I1 Full | FP16 | 64.4 GB | 30.8 min | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | HiDream-I1 Dev | FP16 | 64.4 GB | 11.6 min | Very slow |
| 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.1 Krea [dev] | FP16 | 36.4 GB | 8.2 min | Very slow |
| 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | MOSS-Image SFT | FP16 | 29.3 GB | 9.6 min | Very slow |
| 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.1 Kontext [dev] | FP16 | 37.0 GB | 15.1 min | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.1 [dev] | FP16 | 36.4 GB | 8.2 min | Very slow |
| 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Z-Image | FP16 | 22.7 GB | 11.6 min | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | HiDream-I1 Fast | FP16 | 64.4 GB | 6.8 min | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.1 [schnell] | FP16 | 36.4 GB | 82 s | Slow |
| 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Chroma1-HD | FP16 | 30.0 GB | 13.7 min | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.1 Fill [dev] | FP16 | 36.4 GB | 8.6 min | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Stable Diffusion 3.5 Large Turbo | FP16 | 30.0 GB | 66 s | Slow |
| 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Iris-3B | FP16 | 17.3 GB | 11.0 min | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | CogView4 6B | FP16 | 33.9 GB | 12.3 min | Very slow |
| 54Fair quality (C) · 54 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | SDXL-Lightning (4-step) | FP16 | 8.9 GB | 29 s | Runs well |
| 66Good quality (B) · 66 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Stable Diffusion 3.5 Large | FP16 | 30.0 GB | 7.5 min | Very slow |
| 40Fair quality (C) · 40 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | SDXL Turbo | FP16 | 7.6 GB | 6.2 s | Runs great |
| 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | DreamShaper XL v2 Turbo | FP16 | 8.9 GB | 53 s | Slow |
| 66Good quality (B) · 66 of 100: how capable the model is at FP16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | OmniGen2 | FP16 | 18.0 GB | 11.0 min | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Juggernaut XL v9 | FP16 | 8.9 GB | 3.1 min | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | RealVisXL V5.0 | FP16 | 8.9 GB | 3.1 min | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Kolors | FP16 | 19.8 GB | 2.7 min | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Lumina-Image 2.0 | FP16 | 12.7 GB | 4.8 min | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Animagine XL 4.0 | FP16 | 8.9 GB | 2.9 min | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | Illustrious XL v0.1 | FP16 | 8.9 GB | 2.9 min | Very slow |
| 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Pony Diffusion V6 XL | FP16 | 8.9 GB | 2.6 min | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Playground v2.5 | FP16 | 8.9 GB | 5.1 min | Very slow |
| 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Stable Diffusion 3.5 Medium | FP16 | 18.4 GB | 4.1 min | Very slow |
| 56Good quality (B) · 56 of 100: how capable the model is at FP16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Animagine XL 3.1 | FP16 | 8.9 GB | 2.9 min | Very slow |
| 50Fair quality (C) · 50 of 100: how capable the model is at FP16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | PixArt-Σ XL 1024 | FP16 | 12.8 GB | 62 s | Slow |
| 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | AuraFlow v0.3 | FP16 | 18.5 GB | 9.6 min | Very slow |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Stable Diffusion XL 1.0 | FP16 | 8.9 GB | 3.1 min | Very slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | Stable Diffusion 3 Medium | FP16 | 17.4 GB | 2.1 min | Slow |
| 30Basic quality (D) · 30 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | LCM DreamShaper v7 | FP16 | 2.8 GB | 10 s | Runs great |
| 28Basic quality (D) · 28 of 100: how capable the model is at FP16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | SD Turbo | FP16 | 3.3 GB | 3.3 s | Runs great |
| 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Stable Cascade | FP16 | 14.0 GB | 3.4 min | Very slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Hunyuan-DiT v1.2 | FP16 | 9.3 GB | 4.1 min | Very slow |
| 42Fair quality (C) · 42 of 100: how capable the model is at FP16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Realistic Vision V5.1 | FP16 | 2.8 GB | 49 s | Slow |
| 48Fair quality (C) · 48 of 100: how capable the model is at FP16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | Kandinsky 3 | FP16 | 25.9 GB | 6.2 min | Very slow |
| 40Fair quality (C) · 40 of 100: how capable the model is at FP16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | DreamShaper 8 | FP16 | 2.8 GB | 49 s | Slow |
| 35Basic quality (D) · 35 of 100: how capable the model is at FP16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | Stable Diffusion 1.5 | FP16 | 2.8 GB | 49 s | Slow |
Video models on the AMD EPYC 7002 / 7003 series NAS (512 GB)
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | Cosmos3-Super Image2Video (64B) | FP16 | 139 GB | 25.7 h | Very slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | MiniMax H3 (33B, video + audio) | FP16 | 113 GB | 11.9 h | Very slow |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | LTX-2.5 (22B, video + audio) | FP16 | 82.5 GB | 62.3 min | Very slow |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Kandinsky 6.0 Pro (29B, video + audio) | FP16 | 85.4 GB | 33.6 h | Very slow |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | LTX-2.3 | FP16 | 75.4 GB | 4.6 h | Very slow |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.2 I2V A14B | FP16 | 58.0 GB | 18.3 h | Very slow |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.2 T2V A14B | FP16 | 58.0 GB | 18.3 h | Very slow |
| 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | LTX-2 (19B, video + audio) | FP16 | 75.4 GB | 4.6 h | Very slow |
| 85Excellent quality (A+) · 85 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | Bernini-R A14B (video generation + editing) | FP16 | 46.0 GB | 8.2 h | Very slow |
| 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | HunyuanVideo 1.5 (8.3B) | FP16 | 55.2 GB | 20.5 h | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | SCAIL-2 14B (character animation) | FP16 | 52.5 GB | 13.7 h | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.2 Animate 14B | FP16 | 65.2 GB | 21.7 h | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 VACE 14B | FP16 | 64.6 GB | 25.1 h | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | SkyReels V2 T2V 14B 720P | FP16 | 60.8 GB | 27.4 h | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 I2V 14B 720P | FP16 | 59.3 GB | 22.8 h | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 T2V 14B | FP16 | 58.0 GB | 22.8 h | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Step-Video T2V (30B) | FP16 | 78.2 GB | 35.4 h | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Wan-Dancer 14B (music to dance) | FP16 | 77.0 GB | 50.2 h | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.2 S2V 14B (speech to video) | FP16 | 50.4 GB | 11.4 h | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | HY-OmniWeaving (8.3B, unified video) | FP16 | 41.9 GB | 6.1 h | Very slow |
| 74Great quality (A) · 74 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | FramePack I2V (HunyuanVideo 13B) | FP16 | 49.6 GB | 4.6 h | Very slow |
| 74Great quality (A) · 74 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 I2V 14B 480P | FP16 | 46.5 GB | 10.3 h | Very slow |
| 74Great quality (A) · 74 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | SkyReels V2 T2V 14B 540P | FP16 | 49.0 GB | 16.0 h | Very slow |
| 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | HunyuanVideo 13B | FP16 | 64.8 GB | 27.4 h | Very slow |
| 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | HunyuanVideo I2V 13B | FP16 | 64.8 GB | 27.4 h | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at FP8, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | LTX-Video 13B 0.9.8 (distilled) | FP8 | 30.1 GB | 37.0 min | Slow |
| 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | Cosmos-Predict2 14B Video2World | FP16 | 57.6 GB | 18.3 h | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.2 TI2V 5B | FP16 | 43.1 GB | 5.9 h | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | Open-Sora 2.0 (11B) | FP16 | 43.8 GB | 34.2 h | Very slow |
| 68Good quality (B) · 68 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | Kandinsky 6.0 Video Lite (3B, video + audio) | FP16 | 31.5 GB | 3.5 h | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Mochi 1 Preview (10B) | FP16 | 41.4 GB | 12.6 h | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | Cosmos 1.0 Diffusion 7B Text2World | FP16 | 48.2 GB | 6.3 h | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | CogVideoX 1.5 5B | FP16 | 50.3 GB | 10.3 h | Very slow |
| 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | Cosmos-Predict2 2B Video2World | FP16 | 33.6 GB | 2.9 h | Very slow |
| 50Fair quality (C) · 50 of 100: how capable the model is at FP16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | LTX-Video 2B | FP16 | 17.5 GB | 17.1 min | Slow |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | CogVideoX 5B I2V | FP16 | 25.3 GB | 2.1 h | Very slow |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | CogVideoX 5B | FP16 | 24.1 GB | 2.1 h | Very slow |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 VACE 1.3B | FP16 | 21.5 GB | 3.0 h | Very slow |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 T2V 1.3B | FP16 | 19.8 GB | 2.7 h | Very slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | SkyReels V2 Diffusion Forcing 1.3B | FP16 | 23.8 GB | 3.4 h | Very slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | Allegro (2.8B) | FP16 | 38.2 GB | 13.7 h | Very slow |
| 44Fair quality (C) · 44 of 100: how capable the model is at FP8, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | CogVideoX 2B | FP8 | 11.0 GB | 37.0 min | Slow |
| 35Basic quality (D) · 35 of 100: how capable the model is at FP16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | AnimateDiff v3 (SD 1.5) | FP16 | 4.7 GB | 13.7 min | Slow |
Decision models on the AMD EPYC 7002 / 7003 series NAS (512 GB)
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | OpenAI Privacy Filter | F16 | 3.8 GB | 44 ms | Runs great |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | Llama Prompt Guard 2 22M | Q8_0 | 0.7 GB | 103 ms | Runs great |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | GLiNER2.5 Decide | Q8_0 | 1.2 GB | 345 ms | Runs well |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | Jina Reranker v3.5 | Q8_0 | 39.2 GB | 780 ms | Slow |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | Jina Reranker v3 | Q8_0 | 39.2 GB | 780 ms | Slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Reranker 4B | Q8_0 | 42.8 GB | 5.2 s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | GLiClass Base v3 | Q8_0 | 0.8 GB | 136 ms | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | FRIDA-Decisions | Q8_0 | 1.7 GB | 574 ms | Slow |
| 74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | GLiClass ModernBERT Large v3 | Q8_0 | 1.1 GB | 282 ms | Runs well |
| 64Good quality (B) · 64 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | DeBERTa v3 Base Zero-shot v2.0 | Q8_0 | 0.8 GB | 129 ms | Runs great |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Reranker 8B | Q8_0 | 47.3 GB | 10.6 s | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | GTE Reranker ModernBERT Base | Q8_0 | 0.8 GB | 213 ms | Runs well |
| 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Llama Prompt Guard 2 86M | Q8_0 | 1.0 GB | 394 ms | Runs well |
| 88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | zerank-2 | Q8_0 | 42.8 GB | 5.2 s | Very slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Skywork Reward V2 Llama 3.1 8B | Q8_0 | 10.4 GB | 20.9 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Reranker 0.6B | Q8_0 | 39.2 GB | 780 ms | Slow |
| 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | DeBERTa v3 Large Zero-shot v2.0 | Q8_0 | 1.2 GB | 310 ms | Runs well |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Granite Embedding Reranker English R2 | Q8_0 | 0.8 GB | 213 ms | Runs well |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | GLiNER2 Base v1 | Q8_0 | 0.9 GB | 150 ms | Runs well |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Shieldstral 1.0 3B | Q8_0 | 42.6 GB | 5.0 s | Very slow |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Intern-Decision 4B | Q8_0 | 43.2 GB | 3.0 s | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | ModernBERT Large Zero-shot v2.0 | Q8_0 | 1.1 GB | 282 ms | Runs well |
| 82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | mxbai-rerank-large v2 | Q8_0 | 40.0 GB | 2.0 s | Very slow |
| 82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | d1 3B | Q8_0 | 41.6 GB | 2.0 s | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | mxbai-rerank-base v2 | Q8_0 | 38.8 GB | 638 ms | Slow |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | GLiDE 27B | Q8_0 | 34.8 GB | 18.2 s | Very slow |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | GLiGuard 300M | Q8_0 | 0.9 GB | 296 ms | Runs well |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Laya | Q8_0 | 1.2 GB | 296 ms | Runs well |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | zerank-1 small | Q8_0 | 40.3 GB | 2.2 s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | ProtectAI DeBERTa v3 Prompt Injection v2 | Q8_0 | 0.8 GB | 254 ms | Runs well |
| 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | Jina Reranker v2 Base Multilingual | Q8_0 | 1.0 GB | 394 ms | Runs well |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | Skywork Reward V2 Qwen3 8B | Q8_0 | 10.5 GB | 21.1 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | LemonDecider 4B | Q8_0 | 6.5 GB | 3.2 s | Very slow |
| 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | ms-marco MiniLM L6 v2 | F16 | 0.7 GB | 41 ms | Runs great |
| 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | DuoGuard 0.5B | Q8_0 | 1.2 GB | 686 ms | Slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | BGE Reranker v2 Gemma | Q8_0 | 41.0 GB | 3.3 s | Very slow |
| 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | BGE Reranker v2 M3 | Q8_0 | 1.3 GB | 797 ms | Slow |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3Guard Gen 8B | Q8_0 | 47.3 GB | 11.2 s | Very slow |
| 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Laya Multilingual | Q8_0 | 1.0 GB | 227 ms | Runs well |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3Guard Gen 4B | Q8_0 | 43.3 GB | 6.0 s | Very slow |
| 62Good quality (B) · 62 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | HHEM 2.1 Open | Q8_0 | 0.7 GB | 310 ms | Runs well |
| 62Good quality (B) · 62 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | GLiNER Multi v2.1 | Q8_0 | 1.0 GB | 206 ms | Runs well |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Llama 3.1 Nemotron Safety Guard 8B v3 | Q8_0 | 47.1 GB | 11.0 s | Very slow |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Granite Guardian 4.1 8B | Q8_0 | 47.5 GB | 11.4 s | Very slow |
| 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Skywork Reward V2 Qwen3 0.6B | Q8_0 | 1.4 GB | 1.7 s | Very slow |
| 74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3Guard Gen 0.6B | Q8_0 | 39.3 GB | 1.0 s | Slow |
| 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | NuExtract 2.0 4B | Q8_0 | 42.3 GB | 3.7 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | WildGuard 7B | Q8_0 | 46.2 GB | 9.9 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | Granite Guardian 3.3 8B | Q8_0 | 47.3 GB | 11.2 s | Very slow |
| 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | BART Large MNLI | Q8_0 | 1.1 GB | 289 ms | Runs well |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | CompassJudger-2 32B | Q8_0 | 73.5 GB | 153.0 s | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | NuExtract 2.0 2B | Q8_0 | 40.7 GB | 2.2 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | gpt-oss-safeguard 120B | MXFP4 | 105 GB | 12.5 s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | LettuceDetect Large | Q8_0 | 1.1 GB | 1.1 s | Slow |
| 79Great quality (A) · 79 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | DynaGuard 8B | Q8_0 | 47.3 GB | 28.9 s | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron Content Safety Reasoning 4B | Q8_0 | 43.1 GB | 15.2 s | Very slow |
| 76Great quality (A) · 76 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | gpt-oss-safeguard 20B | MXFP4 | 50.6 GB | 9.0 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | Selene 1 Mini 8B | Q8_0 | 47.1 GB | 37.5 s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | ShieldGemma 27B | Q8_0 | 67.9 GB | 35.2 s | Very slow |
| 64Good quality (B) · 64 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | ShieldGemma 9B | Q8_0 | 48.8 GB | 12.0 s | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | CompassJudger-2 7B | Q8_0 | 46.5 GB | 35.6 s | Very slow |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | Llama Guard 3 8B | Q8_0 | 47.1 GB | 11.0 s | Very slow |
| 58Good quality (B) · 58 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | ShieldGemma 2B | Q8_0 | 41.3 GB | 3.4 s | Very slow |
| 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | DynaGuard 1.7B | Q8_0 | 40.7 GB | 7.2 s | Very slow |
| 74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Prometheus 2 8x7B | Q8_0 | 88.1 GB | 60.8 s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | Llama Guard 4 12B | Q8_0 | 51.4 GB | 16.4 s | Very slow |
| 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Llama Guard 3 1B | Q8_0 | 39.9 GB | 2.1 s | Very slow |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | Prometheus 2 7B | Q8_0 | 46.2 GB | 33.8 s | Very slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | GLIDER | Q8_0 | 43.1 GB | 17.8 s | Very slow |
| 64Good quality (B) · 64 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Flow Judge v0.1 | Q8_0 | 43.1 GB | 17.8 s | Very slow |
Upgrade picks
GPUs that would run more LLMs without offloading, at launch prices. Right now 459 of 619 LLMs run well on your AMD EPYC 7002 / 7003 series NAS (512 GB).
Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.