Local LLMs
LLMs you can run locally, with memory needs per quantization, context length and license.
| Model | Made by | Architecture | Parameters | Max context | Quality tier | Released |
|---|---|---|---|---|---|---|
| MiMo V2.6 Pro (1T-A42B) | Xiaomi | Mixture of experts | 1,024B (42.0B active) | 1,024K | 99 | 2026-09 |
| Kimi K3 (2.8T) | Moonshot AI | Mixture of experts | 2,780B (104.0B active) | 1,024K | 98 | 2026-07 |
| Qwen3.8 Max (2.4T-A95B) | Alibaba | Mixture of experts | 2,446B (95.0B active) | 256K | 98.3 | 2026-08 |
| GLM-5.3 (753B-A40B) | Zhipu AI | Mixture of experts | 753B (40.0B active) | 1,024K | 97.8 | 2026-08 |
| MiMo V2.6 Flash (310B-A15B) | Xiaomi | Mixture of experts | 311B (15.0B active) | 1,024K | 97.2 | 2026-09 |
| DeepSeek V4 Pro (1.6T-A49B) | DeepSeek | Mixture of experts | 1,651B (49.0B active) | 1,024K | 96.8 | 2026-08 |
| GLM-5.2 (753B-A40B) | Zhipu AI | Mixture of experts | 753B (40.0B active) | 1,024K | 96.3 | 2026-06 |
| Hunyuan Hy4 Preview (780B) | Tencent | Mixture of experts | 780B (55.0B active) | 1,024K | 96.3 | 2026-08 |
| Qwen3.8 Flash Next (125B-A6B) | Alibaba | Mixture of experts | 180B (6.0B active) | 256K | 96.5 | 2026-08 |
| DeepSeek V4 Pro (April 2026) | DeepSeek | Mixture of experts | 1,599B (49.0B active) | 1,024K | 95 | 2026-04 |
| GLM-5.1 (744B-A40B) | Zhipu AI | Mixture of experts | 754B (40.0B active) | 198K | 95.2 | 2026-04 |
| GLM-5.3 Flash (320B-A18B) | Zhipu AI | Mixture of experts | 321B (18.0B active) | 1,024K | 95.4 | 2026-08 |
| Kimi K2.6 (1T) | Moonshot AI | Mixture of experts | 1,027B (32.0B active) | 256K | 95.7 | 2026-04 |
| Qwen3.5 397B-A17B | Alibaba | Mixture of experts | 403B (17.0B active) | 256K | 95 | 2026-02 |
| Qwen3.8 27B | Alibaba | Dense | 28B | 256K | 95.6 | 2026-08 |
| DeepSeek V4 Flash (284B-A13B) | DeepSeek | Mixture of experts | 304B (13.0B active) | 1,024K | 94 | 2026-07 |
| GLM-5 (744B-A40B) | Zhipu AI | Mixture of experts | 754B (40.0B active) | 198K | 94.7 | 2026-02 |
| Hunyuan Hy3 (295B-A21B) | Tencent | Mixture of experts | 299B (21.0B active) | 256K | 94.4 | 2026-07 |
| Kimi K2.5 (1T) | Moonshot AI | Mixture of experts | 1,027B (32.0B active) | 256K | 94.3 | 2026-01 |
| Kimi K2.7 Code (1T) | Moonshot AI | Mixture of experts | 1,027B (32.0B active) | 256K | 94.3 | 2026-06 |
| MiniMax M3 (427B) | MiniMax | Mixture of experts | 427B (25.0B active) | 1,024K | 94.8 | 2026-06 |
| Ornith 1.5 397B-A17B | DeepReinforce | Mixture of experts | 403B (17.0B active) | 256K | 94 | 2026-08 |
| Qwen3.6 27B | Alibaba | Dense | 28B | 256K | 94 | 2026-04 |
| DeepSeek R1 0528 (671B) | DeepSeek | Mixture of experts | 685B (37.0B active) | 160K | 93 | 2025-05 |
| DeepSeek V3.2 (671B) | DeepSeek | Mixture of experts | 685B (37.0B active) | 160K | 93.7 | 2025-12 |
| GLM-4.7 (355B-A32B) | Zhipu AI | Mixture of experts | 358B (32.0B active) | 198K | 93.5 | 2025-12 |
| Kimi K2 Thinking (1T) | Moonshot AI | Mixture of experts | 1,026B (32.0B active) | 256K | 93 | 2025-11 |
| MiniMax M2.7 (230B-A10B) | MiniMax | Mixture of experts | 229B (10.0B active) | 200K | 93.5 | 2026-04 |
| Nemotron 3 Ultra 550B-A55B | NVIDIA | Mixture of experts | 561B (55.0B active) | 256K | 93.2 | 2026-06 |
| Qwen3.5 27B | Alibaba | Dense | 28B | 256K | 93 | 2026-02 |
| Qwen3.5 35B-A3B | Alibaba | Mixture of experts | 36B (3.0B active) | 256K | 93 | 2026-02 |
| Qwen3.6 35B-A3B | Alibaba | Mixture of experts | 36B (3.0B active) | 256K | 93.5 | 2026-04 |
| DeepSeek R1 (671B) | DeepSeek | Mixture of experts | 671B (37.0B active) | 128K | 92 | 2025-01 |
| DeepSeek V3.2 Exp | DeepSeek | Mixture of experts | 685B (37.0B active) | 160K | 92.5 | 2025-09 |
| DeepSeek V4 Flash (April 2026) | DeepSeek | Mixture of experts | 291B (13.0B active) | 1,024K | 92 | 2026-04 |
| Gemma 4 26B-A4B | Mixture of experts | 26B (3.8B active) | 256K | 92.5 | 2026-03 | |
| MiniMax M2.5 (230B-A10B) | MiniMax | Mixture of experts | 229B (10.0B active) | 192K | 92.5 | 2026-02 |
| Qwen3.5 122B-A10B | Alibaba | Mixture of experts | 125B (10.0B active) | 256K | 92 | 2026-02 |
| Step 3.5 Flash (196B-A11B) | StepFun | Mixture of experts | 199B (11.0B active) | 256K | 92 | 2026-02 |
| DeepSeek V3.1 (671B) | DeepSeek | Mixture of experts | 685B (37.0B active) | 160K | 91 | 2025-08 |
| DeepSeek V3.1 Terminus | DeepSeek | Mixture of experts | 685B (37.0B active) | 160K | 91.5 | 2025-09 |
| Gemma 4 31B | Dense | 31B | 256K | 91 | 2026-03 | |
| Kimi K2 Instruct 0905 (1T) | Moonshot AI | Mixture of experts | 1,026B (32.0B active) | 256K | 91 | 2025-09 |
| MiniMax M2.1 (230B-A10B) | MiniMax | Mixture of experts | 229B (10.0B active) | 192K | 91 | 2025-12 |
| Qwen3 235B-A22B Thinking 2507 | Alibaba | Mixture of experts | 235B (22.0B active) | 256K | 91 | 2025-07 |
| Qwen3 Coder 480B-A35B | Alibaba | Mixture of experts | 480B (35.0B active) | 256K | 91 | 2025-07 |
| Ling 1T | Ant Group (inclusionAI) | Mixture of experts | 1,000B (50.0B active) | 32K | 90 | 2025-10 |
| MiMo V2 Flash (309B-A15B) | Xiaomi | Mixture of experts | 310B (15.0B active) | 256K | 90.9 | 2025-12 |
| Mistral Large 3 (675B) | Mistral AI | Mixture of experts | 675B (41.0B active) | 288K | 90 | 2025-12 |
| Ornith 1.5 35B-A3B | DeepReinforce | Mixture of experts | 36B (3.0B active) | 256K | 90 | 2026-08 |
| Qwen3 235B-A22B Instruct 2507 | Alibaba | Mixture of experts | 235B (22.0B active) | 256K | 90 | 2025-07 |
| Solar Open 2 250B | Upstage | Mixture of experts | 250B (15.0B active) | 1,024K | 90 | 2026-07 |
| TMax 27B | Ai2 | Dense | 27B | 256K | 90 | 2026-06 |
| Apodex 1.1 Mini (35B-A3B) | Apodex | Mixture of experts | 36B (3.0B active) | 256K | 89 | 2026-08 |
| Kimi K2 Instruct (July 2025) | Moonshot AI | Mixture of experts | 1,026B (32.0B active) | 128K | 89 | 2025-07 |
| MiniMax M2 (230B-A10B) | MiniMax | Mixture of experts | 229B (10.0B active) | 192K | 89 | 2025-10 |
| Ornith 1.0 35B-A3B | DeepReinforce | Mixture of experts | 36B (3.0B active) | 256K | 89 | 2026-06 |
| Qwen3-VL 235B-A22B Instruct | Alibaba | Mixture of experts | 236B (22.0B active) | 256K | 89 | 2025-09 |
| Fara 1.5 27B | Microsoft | Dense | 27B | 256K | 88 | 2026-07 |
| GLM-4.7 Flash (30B-A3B) | Zhipu AI | Mixture of experts | 31B (3.0B active) | 198K | 88 | 2026-01 |
| K-EXAONE 236B-A23B | LG AI Research | Mixture of experts | 237B (23.0B active) | 256K | 88 | 2025-12 |
| LongCat Flash Chat (560B) | Meituan | Mixture of experts | 562B (27.0B active) | 128K | 88 | 2025-08 |
| Nemotron 3 Super 120B-A12B | NVIDIA | Mixture of experts | 124B (12.0B active) | 256K | 88 | 2026-03 |
| Qwen3 235B-A22B | Alibaba | Mixture of experts | 235B (22.0B active) | 32K | 88 | 2025-04 |
| Agents-A1 35B-A3B | InternScience | Mixture of experts | 35B (3.0B active) | 256K | 87 | 2026-06 |
| Qwen AgentWorld 35B-A3B | Alibaba | Mixture of experts | 35B (3.0B active) | 256K | 87 | 2026-06 |
| Devstral 2 (123B) | Mistral AI | Dense | 125B | 256K | 86 | 2025-12 |
| InternVL3.5 241B-A28B | Shanghai AI Lab | Mixture of experts | 241B (28.0B active) | 40K | 86 | 2025-08 |
| Laguna S 2.1 (118B-A8B) | Poolside | Mixture of experts | 118B (8.0B active) | 1,024K | 86 | 2026-07 |
| Llama 3.1 Nemotron Ultra 253B v1 | NVIDIA | Dense | 253B | 128K | 86 | 2025-04 |
| Qwen3 Coder Next (80B-A3B) | Alibaba | Mixture of experts | 80B (3.0B active) | 256K | 86 | 2026-01 |
| Qwen3-Next 80B-A3B Thinking | Alibaba | Mixture of experts | 81B (3.0B active) | 256K | 86 | 2025-09 |
| UI-Mate 27B | Tencent | Dense | 27B | 256K | 86 | 2026-08 |
| Baichuan M3 235B-A22B | Baichuan | Mixture of experts | 235B (22.0B active) | 40K | 84 | 2026-01 |
| GLM-4.6 (355B-A32B) | Zhipu AI | Mixture of experts | 357B (32.0B active) | 198K | 84 | 2025-09 |
| gpt-oss 120B | OpenAI | Mixture of experts | 117B (5.1B active) | 128K | 84 | 2025-08 |
| Qwen3 Next 80B-A3B Instruct | Alibaba | Mixture of experts | 81B (3.0B active) | 256K | 84 | 2025-09 |
| DeepSeek V3 0324 (671B) | DeepSeek | Mixture of experts | 685B (37.0B active) | 160K | 83 | 2025-03 |
| Gemma 4 12B | Dense | 12B | 256K | 83 | 2026-05 | |
| GLM-4.6V | Zhipu AI | Mixture of experts | 108B (12.0B active) | 128K | 83 | 2025-12 |
| MiniMax-M1 80K | MiniMax | Mixture of experts | 456B (45.9B active) | 976K | 83 | 2025-06 |
| Command A Reasoning (111B) | Cohere | Dense | 111B | 256K | 82 | 2025-08 |
| EXAONE 4.5 33B | LG AI Research | Dense | 34B | 256K | 82 | 2026-04 |
| GLM-4.5 (355B-A32B) | Zhipu AI | Mixture of experts | 358B (32.0B active) | 128K | 82 | 2025-07 |
| Laguna XS 2.1 (33B-A3B) | Poolside | Mixture of experts | 33B (3.0B active) | 256K | 82 | 2026-06 |
| Ling 3.0 Flash (127B-A5B) | Ant Group (inclusionAI) | Mixture of experts | 127B (5.1B active) | 256K | 82 | 2026-08 |
| Llama 3.3 70B Instruct | Meta | Dense | 71B | 128K | 82 | 2024-12 |
| MiniMax-M1 40K | MiniMax | Mixture of experts | 456B (45.9B active) | 976K | 82 | 2025-06 |
| Command A (111B) | Cohere | Dense | 111B | 256K | 80 | 2025-03 |
| ERNIE 4.5 300B-A47B | Baidu | Mixture of experts | 300B (47.0B active) | 128K | 80 | 2025-06 |
| GLM-4.5V | Zhipu AI | Mixture of experts | 108B (12.0B active) | 64K | 80 | 2025-08 |
| Hermes 4 405B | Nous Research | Dense | 406B | 128K | 80 | 2025-08 |
| Kimi Dev 72B | Moonshot AI | Dense | 73B | 128K | 80 | 2025-06 |
| Llama 3.3 Nemotron Super 49B v1.5 | NVIDIA | Dense | 50B | 128K | 80 | 2025-07 |
| Mistral Large 2.1 (2411) | Mistral AI | Dense | 123B | 128K | 80 | 2024-11 |
| Mistral Medium 3.5 (128B) | Mistral AI | Dense | 128B | 256K | 80.4 | 2026-03 |
| Mistral Small 4 (119B-A6B) | Mistral AI | Mixture of experts | 119B (6.0B active) | 1,024K | 80.6 | 2026-03 |
| Nemotron Cascade 2 30B-A3B | NVIDIA | Mixture of experts | 32B (3.5B active) | 256K | 80 | 2026-03 |
| North Mini Code 1.0 (30B-A3B) | Cohere | Mixture of experts | 30B (3.0B active) | 488K | 80 | 2026-06 |
| Seed-OSS 36B Instruct | ByteDance | Dense | 36B | 512K | 80 | 2025-08 |
| Solar Open 100B | Upstage | Mixture of experts | 103B (12.0B active) | 128K | 80 | 2025-12 |
| Trinity Large Thinking (400B-A13B) | Arcee AI | Mixture of experts | 399B (13.0B active) | 256K | 80.3 | 2026-04 |
| Hermes 4.3 36B | Nous Research | Dense | 36B | 512K | 79 | 2025-11 |
| DiffusionGemma 26B-A4B IT | Mixture of experts | 26B (4.0B active) | 256K | 78 | 2026-06 | |
| GLM-4.5 Air (106B-A12B) | Zhipu AI | Mixture of experts | 110B (12.0B active) | 128K | 78 | 2025-07 |
| Mistral Large 2 (123B) | Mistral AI | Dense | 123B | 128K | 78 | 2024-07 |
| Pixtral Large 124B (2411) | Mistral AI | Dense | 124B | 128K | 78 | 2024-11 |
| Qwen3 30B-A3B Thinking 2507 | Alibaba | Mixture of experts | 31B (3.3B active) | 256K | 78 | 2025-07 |
| Qwen3 Coder 30B-A3B | Alibaba | Mixture of experts | 31B (3.3B active) | 256K | 78 | 2025-07 |
| Qwen3-VL 32B Instruct | Alibaba | Dense | 33B | 256K | 78 | 2025-10 |
| QwQ 32B | Alibaba | Dense | 33B | 40K | 78 | 2025-03 |
| Ring Flash 2.0 (100B-A6B) | Ant Group (inclusionAI) | Mixture of experts | 103B (6.1B active) | 32K | 78 | 2025-09 |
| Nemotron 3 Nano 30B-A3B | NVIDIA | Mixture of experts | 32B (3.5B active) | 256K | 77 | 2025-12 |
| Ornith 1.5 9B | DeepReinforce | Dense | 9.7B | 256K | 77 | 2026-08 |
| Qwen3 30B-A3B Instruct 2507 | Alibaba | Mixture of experts | 31B (3.3B active) | 256K | 77 | 2025-07 |
| Devstral Small 2 (24B) | Mistral AI | Dense | 24B | 384K | 76 | 2025-12 |
| dots.llm1 Instruct | rednote | Mixture of experts | 143B (14.0B active) | 32K | 76 | 2025-05 |
| Granite 4.2 30B | IBM | Dense | 29B | 128K | 76 | 2026-08 |
| InternVL3.5 38B | Shanghai AI Lab | Dense | 38B | 40K | 76 | 2025-08 |
| Ling Flash 2.0 (100B-A6B) | Ant Group (inclusionAI) | Mixture of experts | 103B (6.1B active) | 32K | 76 | 2025-09 |
| MiniMax-Text-01 | MiniMax | Mixture of experts | 456B (45.9B active) | 976K | 76 | 2025-01 |
| Olmo 3.1 32B Think | Ai2 | Dense | 32B | 64K | 76 | 2025-12 |
| OpenReasoning Nemotron 32B | NVIDIA | Dense | 33B | 128K | 76 | 2025-07 |
| Qwen3 32B | Alibaba | Dense | 33B | 32K | 76 | 2025-04 |
| Qwen3-VL 30B-A3B Thinking | Alibaba | Mixture of experts | 31B (3.3B active) | 256K | 76 | 2025-09 |
| Qwen3.5 9B | Alibaba | Dense | 9.7B | 256K | 76 | 2026-03 |
| Skywork OR1 32B | Skywork | Dense | 33B | 128K | 76 | 2025-05 |
| Tülu 3 405B | Ai2 | Dense | 406B | 128K | 76 | 2025-01 |
| DeepSeek V3 (671B) | DeepSeek | Mixture of experts | 671B (37.0B active) | 128K | 75 | 2024-12 |
| Ornith 1.0 9B | DeepReinforce | Dense | 9.7B | 256K | 75 | 2026-06 |
| Sarvam 105B | Sarvam AI | Mixture of experts | 106B (10.0B active) | 128K | 75 | 2026-03 |
| DeepSeek Coder V2 (236B) | DeepSeek | Mixture of experts | 236B (21.0B active) | 160K | 74 | 2024-07 |
| EXAONE Deep 32B | LG AI Research | Dense | 32B | 32K | 74 | 2025-03 |
| GLM-4 32B 0414 | Zhipu AI | Dense | 33B | 32K | 74 | 2025-04 |
| GLM-Z1 32B (0414) | Zhipu AI | Dense | 33B | 32K | 74 | 2025-04 |
| Hermes 4 70B | Nous Research | Dense | 71B | 128K | 74 | 2025-08 |
| Hunyuan A13B (80B-A13B) | Tencent | Mixture of experts | 80B (13.0B active) | 32K | 74 | 2025-06 |
| InternVL3 78B | Shanghai AI Lab | Dense | 78B | 32K | 74 | 2025-04 |
| Kimi Linear 48B-A3B | Moonshot AI | Mixture of experts | 49B (3.0B active) | 1,024K | 74 | 2025-10 |
| Leanstral 119B-A6B | Mistral AI | Mixture of experts | 119B (6.5B active) | 256K | 74 | 2026-03 |
| Qwen3 30B-A3B | Alibaba | Mixture of experts | 31B (3.3B active) | 32K | 74 | 2025-04 |
| Qwen3-VL 30B-A3B Instruct | Alibaba | Mixture of experts | 31B (3.3B active) | 256K | 74 | 2025-09 |
| TMax 9B | Ai2 | Dense | 9.0B | 256K | 74 | 2026-06 |
| Qwen2.5 72B Instruct | Alibaba | Dense | 73B | 32K | 73 | 2024-09 |
| Qwen2.5-VL 72B Instruct | Alibaba | Dense | 73B | 125K | 73 | 2025-01 |
| Apriel 1.5 15B Thinker | ServiceNow | Dense | 15B | 128K | 72 | 2025-09 |
| DeepSeek Coder V2 Instruct (0614) | DeepSeek | Mixture of experts | 236B (21.0B active) | 160K | 72 | 2024-06 |
| DeepSeek R1 Distill Llama 70B | DeepSeek | Dense | 71B | 128K | 72 | 2025-01 |
| DeepSeek R1 Distill Qwen 32B | DeepSeek | Dense | 33B | 128K | 72 | 2025-01 |
| DeepSeek V2.5 1210 | DeepSeek | Mixture of experts | 236B (21.0B active) | 160K | 72 | 2024-12 |
| EXAONE 4.0 32B | LG AI Research | Dense | 32B | 128K | 72 | 2025-07 |
| Gemma 3 27B IT | Dense | 27B | 128K | 72 | 2025-03 | |
| Granite 4.1 30B | IBM | Dense | 29B | 128K | 72 | 2026-04 |
| Llama 3.1 405B Instruct | Meta | Dense | 406B | 128K | 72 | 2024-07 |
| Llama 4 Maverick (17B x 128E) | Meta | Mixture of experts | 402B (17.0B active) | 1,024K | 72 | 2025-04 |
| Llama 4 Scout 17B-16E | Meta | Mixture of experts | 109B (17.0B active) | 10,240K | 72 | 2025-04 |
| Magistral Small 1.2 (24B) | Mistral AI | Dense | 24B | 128K | 72 | 2025-09 |
| Nemotron 3.5 Lightning 30B-A3B | NVIDIA | Mixture of experts | 32B (3.0B active) | 256K | 72 | 2026-08 |
| Olmo 3.1 32B Instruct | Ai2 | Dense | 32B | 64K | 72 | 2025-12 |
| Phi-4 Reasoning Plus (14B) | Microsoft | Dense | 15B | 32K | 72 | 2025-04 |
| Tülu 3 70B | Ai2 | Dense | 71B | 128K | 72 | 2024-11 |
| Baichuan M2 32B | Baichuan | Dense | 33B | 128K | 70 | 2025-08 |
| Command R+ (08-2024) 104B | Cohere | Dense | 104B | 128K | 70 | 2024-08 |
| DeepSeek V2.5 | DeepSeek | Mixture of experts | 236B (21.0B active) | 160K | 70 | 2024-09 |
| Devstral Small 1.1 (24B) | Mistral AI | Dense | 24B | 128K | 70 | 2025-07 |
| ERNIE 4.5 21B-A3B Thinking | Baidu | Mixture of experts | 22B (3.0B active) | 128K | 70 | 2025-09 |
| gpt-oss 20B | OpenAI | Mixture of experts | 21B (3.6B active) | 128K | 70 | 2025-08 |
| InternVL3 38B | Shanghai AI Lab | Dense | 38B | 32K | 70 | 2025-04 |
| Jamba 1.5 Large | AI21 Labs | Mixture of experts | 399B (94.0B active) | 256K | 70 | 2024-08 |
| Llama 3.1 70B Instruct | Meta | Dense | 71B | 128K | 70 | 2024-07 |
| Magistral Small 2507 | Mistral AI | Dense | 24B | 40K | 70 | 2025-07 |
| MedGemma 27B Text IT | Dense | 27B | 128K | 70 | 2025-05 | |
| OpenHands LM 32B | All Hands AI | Dense | 33B | 32K | 70 | 2025-03 |
| Phi-4 Reasoning (14B) | Microsoft | Dense | 15B | 32K | 70 | 2025-04 |
| Phi-4 Reasoning Vision 15B | Microsoft | Dense | 15B | 32K | 70 | 2026-01 |
| Qwen2.5 32B Instruct | Alibaba | Dense | 33B | 32K | 70 | 2024-09 |
| Qwen3-Omni 30B-A3B Instruct | Alibaba | Mixture of experts | 35B (3.3B active) | 64K | 70 | 2025-09 |
| Sarvam 30B | Sarvam AI | Mixture of experts | 32B (2.4B active) | 128K | 70 | 2026-03 |
| UI-Mate 9B | Tencent | Dense | 9.4B | 256K | 70 | 2026-08 |
| UI-Venus 2 9B | Ant Group (inclusionAI) | Dense | 9.4B | 256K | 70 | 2026-08 |
| Gemma 4 E4B | Dense | 8.0B | 128K | 68 | 2026-03 | |
| Granite 4.2 8B | IBM | Dense | 8.8B | 128K | 68 | 2026-08 |
| Hermes 3 (Llama 3.1 70B) | Nous Research | Dense | 71B | 128K | 68 | 2024-08 |
| Magistral Small 2506 | Mistral AI | Dense | 24B | 40K | 68 | 2025-06 |
| Mistral Small 3.2 24B | Mistral AI | Dense | 24B | 128K | 68 | 2025-06 |
| QVQ 72B Preview | Alibaba | Dense | 73B | 32K | 68 | 2024-12 |
| Qwen2.5-VL 32B Instruct | Alibaba | Dense | 33B | 125K | 68 | 2025-03 |
| Qwen3 14B | Alibaba | Dense | 15B | 32K | 68 | 2025-04 |
| Qwen2.5 Coder 32B Instruct | Alibaba | Dense | 33B | 32K | 67 | 2024-11 |
| AceReason Nemotron 14B | NVIDIA | Dense | 15B | 128K | 66 | 2025-05 |
| Apertus v1.5 70B | Swiss AI | Dense | 72B | 256K | 66 | 2026-07 |
| Apriel Nemotron 15B Thinker | ServiceNow | Dense | 15B | 64K | 66 | 2025-05 |
| Devstral Small 2505 | Mistral AI | Dense | 24B | 128K | 66 | 2025-05 |
| Hermes 4 14B | Nous Research | Dense | 15B | 40K | 66 | 2025-08 |
| Llama 3.1 Nemotron 70B Instruct | NVIDIA | Dense | 71B | 128K | 66 | 2024-10 |
| Llama 3.2 90B Vision Instruct | Meta | Dense | 89B | 128K | 66 | 2024-09 |
| Ministral 3 14B Reasoning | Mistral AI | Dense | 14B | 256K | 66 | 2025-12 |
| Mistral Small 3.1 24B | Mistral AI | Dense | 24B | 128K | 66 | 2025-03 |
| OpenReasoning Nemotron 14B | NVIDIA | Dense | 15B | 128K | 66 | 2025-07 |
| Qwen2.5 Coder 14B Instruct | Alibaba | Dense | 15B | 32K | 66 | 2024-11 |
| QwQ 32B Preview | Alibaba | Dense | 33B | 32K | 66 | 2024-11 |
| Reka Flash 3.1 (21B) | Reka AI | Dense | 21B | 96K | 66 | 2025-07 |
| Trinity Mini (26B-A3B) | Arcee AI | Mixture of experts | 26B (3.0B active) | 128K | 66 | 2025-12 |
| Command R+ 104B (04-2024) | Cohere | Dense | 104B | 128K | 64 | 2024-04 |
| DeepCoder 14B Preview | Agentica | Dense | 15B | 128K | 64 | 2025-04 |
| ERNIE 4.5 21B-A3B | Baidu | Mixture of experts | 22B (3.0B active) | 128K | 64 | 2025-06 |
| Falcon-H1 34B Instruct | TII | Dense | 34B | 256K | 64 | 2025-05 |
| InternVL3 14B | Shanghai AI Lab | Dense | 15B | 32K | 64 | 2025-04 |
| Ministral 3 14B | Mistral AI | Dense | 14B | 256K | 64 | 2025-12 |
| Mistral Small 3 24B (2501) | Mistral AI | Dense | 24B | 32K | 64 | 2025-01 |
| Phi-4 (14B) | Microsoft | Dense | 15B | 16K | 64 | 2024-12 |
| Qwen2 72B Instruct | Alibaba | Dense | 73B | 32K | 64 | 2024-05 |
| Qwen2-VL 72B Instruct | Alibaba | Dense | 73B | 32K | 64 | 2024-09 |
| Codestral 22B | Mistral AI | Dense | 22B | 32K | 62 | 2024-05 |
| Command R (08-2024) 32B | Cohere | Dense | 32B | 128K | 62 | 2024-08 |
| DeepSeek R1 0528 Qwen3 8B | DeepSeek | Dense | 8.2B | 128K | 62 | 2025-05 |
| DeepSeek R1 Distill Qwen 14B | DeepSeek | Dense | 15B | 128K | 62 | 2025-01 |
| DeepSeek V2 Chat | DeepSeek | Mixture of experts | 236B (21.0B active) | 160K | 62 | 2024-04 |
| Dolphin Mistral 24B Venice Edition | Cognitive Computations | Dense | 24B | 128K | 62 | 2025-06 |
| Gemma 2 27B IT | Dense | 27B | 8K | 62 | 2024-06 | |
| Gemma 4 E2B | Dense | 5.1B | 128K | 62 | 2026-03 | |
| Granite 4.0 H Small (32B-A9B) | IBM | Mixture of experts | 32B (9.0B active) | 128K | 62 | 2025-09 |
| InternVL3.5 8B | Shanghai AI Lab | Dense | 8.5B | 40K | 62 | 2025-08 |
| JanusCoder 14B | Shanghai AI Lab | Dense | 15B | 32K | 62 | 2025-10 |
| Qwen2.5 14B Instruct | Alibaba | Dense | 15B | 32K | 62 | 2024-09 |
| Qwen3-VL 8B Instruct | Alibaba | Dense | 8.8B | 256K | 62 | 2025-10 |
| Qwen3.5 4B | Alibaba | Dense | 4.7B | 256K | 62 | 2026-03 |
| Voxtral Small 24B | Mistral AI | Dense | 24B | 32K | 62 | 2025-07 |
| TMax 4B | Ai2 | Dense | 4.2B | 256K | 61 | 2026-06 |
| Agents-A1 4B | InternScience | Dense | 4.5B | 256K | 60 | 2026-07 |
| Apertus 70B Instruct | Swiss AI | Dense | 71B | 64K | 60 | 2025-09 |
| Aya Expanse 32B | Cohere | Dense | 32B | 128K | 60 | 2024-10 |
| Aya Vision 32B | Cohere | Dense | 33B | 16K | 60 | 2025-03 |
| Dolphin 3.0 R1 Mistral 24B | Cognitive Computations | Dense | 24B | 32K | 60 | 2025-02 |
| Gemma 3 12B IT | Dense | 12B | 128K | 60 | 2025-03 | |
| GLM-4.1V 9B Thinking | Zhipu AI | Dense | 10B | 64K | 60 | 2025-06 |
| Granite 4.1 8B | IBM | Dense | 8.8B | 128K | 60 | 2026-04 |
| Kimi-VL A3B Thinking 2506 | Moonshot AI | Mixture of experts | 16B (3.0B active) | 128K | 60 | 2025-06 |
| Ling 3.0 Tiny (8B-A1.3B) | Ant Group (inclusionAI) | Mixture of experts | 7.9B (1.3B active) | 128K | 60 | 2026-08 |
| MiniCPM-V 4.5 8B | OpenBMB | Dense | 8.7B | 40K | 60 | 2025-08 |
| Nemotron Nano 9B v2 | NVIDIA | Dense | 8.9B | 128K | 60 | 2025-08 |
| OLMo 2 32B Instruct | Ai2 | Dense | 32B | 4K | 60 | 2025-03 |
| Qwen2.5 14B Instruct 1M | Alibaba | Dense | 15B | 986K | 60 | 2025-01 |
| Qwen3 8B | Alibaba | Dense | 8.2B | 32K | 60 | 2025-04 |
| Solar Pro Preview 22B | Upstage | Dense | 22B | 4K | 60 | 2024-09 |
| Fara 1.5 4B | Microsoft | Dense | 4.5B | 256K | 58 | 2026-07 |
| GLM-Z1 9B (0414) | Zhipu AI | Dense | 9.4B | 32K | 58 | 2025-04 |
| Llama 3 70B Instruct | Meta | Dense | 71B | 8K | 58 | 2024-04 |
| MiMo-VL 7B RL | Xiaomi | Dense | 8.3B | 125K | 58 | 2025-05 |
| Qwen2.5-Math 72B Instruct | Alibaba | Dense | 73B | 4K | 58 | 2024-09 |
| Seed-Coder 8B Reasoning | ByteDance | Dense | 8.3B | 64K | 58 | 2025-04 |
| EXAONE Deep 7.8B | LG AI Research | Dense | 7.8B | 32K | 56 | 2025-03 |
| Falcon-H1R 7B | TII | Dense | 7.6B | 256K | 56 | 2025-10 |
| InternVL3 8B | Shanghai AI Lab | Dense | 7.9B | 32K | 56 | 2025-04 |
| LFM2 24B-A2B | Liquid AI | Mixture of experts | 24B (2.3B active) | 125K | 56 | 2026-02 |
| Mage-VL 4.7B | Microsoft | Dense | 4.7B | 256K | 56 | 2026-07 |
| MiMo 7B RL | Xiaomi | Dense | 7.8B | 32K | 56 | 2025-04 |
| Ministral 3 8B | Mistral AI | Dense | 8.9B | 256K | 56 | 2025-12 |
| Phi-3.5 MoE (42B-A6.6B) | Microsoft | Mixture of experts | 42B (6.6B active) | 128K | 56 | 2024-08 |
| Qwen3 4B Thinking 2507 | Alibaba | Dense | 4.0B | 256K | 56 | 2025-08 |
| Seed-Coder 8B Instruct | ByteDance | Dense | 8.3B | 32K | 56 | 2025-04 |
| Mixtral 8x22B Instruct | Mistral AI | Mixture of experts | 141B (39.0B active) | 64K | 55 | 2024-04 |
| Qwen2.5 Coder 7B Instruct | Alibaba | Dense | 7.6B | 32K | 55 | 2024-11 |
| Command R 35B (v01) | Cohere | Dense | 35B | 128K | 54 | 2024-03 |
| Llama 3.1 Nemotron Nano 8B v1 | NVIDIA | Dense | 8.0B | 128K | 54 | 2025-03 |
| MiniCPM4 8B | OpenBMB | Dense | 8.2B | 32K | 54 | 2025-06 |
| Mistral Small 22B (2409) | Mistral AI | Dense | 22B | 32K | 54 | 2024-09 |
| Olmo 3 7B Think | Ai2 | Dense | 7.3B | 64K | 54 | 2025-11 |
| OpenReasoning Nemotron 7B | NVIDIA | Dense | 7.6B | 128K | 54 | 2025-07 |
| Qwen3 4B Instruct 2507 | Alibaba | Dense | 4.0B | 256K | 54 | 2025-08 |
| Qwen3-VL 4B Instruct | Alibaba | Dense | 4.4B | 256K | 54 | 2025-10 |
| DeepSeek Coder V2 Lite (16B) | DeepSeek | Mixture of experts | 16B (2.4B active) | 160K | 52 | 2024-06 |
| Falcon 3 10B Instruct | TII | Dense | 10B | 32K | 52 | 2024-12 |
| GLM-4 9B 0414 | Zhipu AI | Dense | 9.4B | 32K | 52 | 2025-04 |
| Granite 4.2 3B | IBM | Dense | 3.7B | 128K | 52 | 2026-08 |
| Hunyuan 7B Instruct | Tencent | Dense | 7.5B | 32K | 52 | 2025-07 |
| LFM2.5 8B-A1B | Liquid AI | Mixture of experts | 8.5B (1.5B active) | 125K | 52 | 2026-05 |
| Ling Mini 2.0 (16B-A1.4B) | Ant Group (inclusionAI) | Mixture of experts | 16B (1.4B active) | 32K | 52 | 2025-09 |
| MiniCPM-o 2.6 8B | OpenBMB | Dense | 8.7B | 32K | 52 | 2025-01 |
| Mistral Nemo 12B Instruct | Mistral AI | Dense | 12B | 128K | 52 | 2024-07 |
| Nous Hermes 2 Mixtral 8x7B DPO | Nous Research | Mixture of experts | 47B (12.9B active) | 32K | 52 | 2024-01 |
| Qwen1.5 110B Chat | Alibaba | Dense | 111B | 32K | 52 | 2024-04 |
| Qwen2.5 7B Instruct | Alibaba | Dense | 7.6B | 32K | 52 | 2024-09 |
| Qwen2.5-VL 7B Instruct | Alibaba | Dense | 8.3B | 125K | 52 | 2025-01 |
| Aya 23 35B | Cohere | Dense | 35B | 8K | 50 | 2024-05 |
| DeepSeek R1 Distill Llama 8B | DeepSeek | Dense | 8.0B | 128K | 50 | 2025-01 |
| Falcon-H1 7B Instruct | TII | Dense | 7.6B | 256K | 50 | 2025-05 |
| Fara 7B | Microsoft | Dense | 8.3B | 125K | 50 | 2025-10 |
| Gemma 2 9B IT | Dense | 9.2B | 8K | 50 | 2024-06 | |
| Grok-1 (314B) | xAI | Mixture of experts | 314B (86.0B active) | 8K | 50 | 2024-03 |
| InternLM2.5 20B Chat | Shanghai AI Lab | Dense | 20B | 32K | 50 | 2024-07 |
| InternLM3 8B Instruct | Shanghai AI Lab | Dense | 8.8B | 32K | 50 | 2025-01 |
| InternVL2.5 8B | Shanghai AI Lab | Dense | 8.1B | 32K | 50 | 2024-11 |
| Jamba 1.5 Mini | AI21 Labs | Mixture of experts | 52B (12.0B active) | 256K | 50 | 2024-08 |
| Kimi-VL A3B Instruct | Moonshot AI | Mixture of experts | 16B (3.0B active) | 128K | 50 | 2025-04 |
| Mixtral 8x7B Instruct | Mistral AI | Mixture of experts | 47B (12.9B active) | 32K | 50 | 2023-12 |
| Nemotron 3 Nano 4B | NVIDIA | Dense | 4.0B | 256K | 50 | 2026-03 |
| Olmo 3 7B Instruct | Ai2 | Dense | 7.3B | 64K | 50 | 2025-11 |
| Pixtral 12B (2409) | Mistral AI | Dense | 12B | 128K | 50 | 2024-09 |
| Qwen2 57B-A14B Instruct | Alibaba | Mixture of experts | 57B (14.0B active) | 32K | 50 | 2024-06 |
| Qwen2.5 7B Instruct 1M | Alibaba | Dense | 7.6B | 986K | 50 | 2025-01 |
| Qwen3 4B | Alibaba | Dense | 4.0B | 32K | 50 | 2025-04 |
| Snowflake Arctic Instruct | Snowflake | Mixture of experts | 479B (17.0B active) | 4K | 50 | 2024-04 |
| Yi 1.5 34B Chat | 01.AI | Dense | 34B | 4K | 50 | 2024-05 |
| GLM-4 9B Chat (2024) | Zhipu AI | Dense | 9.4B | 128K | 48 | 2024-06 |
| GLM-4 9B Chat 1M | Zhipu AI | Dense | 9.5B | 1,024K | 48 | 2024-06 |
| Hermes 3 (Llama 3.1 8B) | Nous Research | Dense | 8.0B | 128K | 48 | 2024-08 |
| Llama 3.1 8B Instruct | Meta | Dense | 8.0B | 128K | 48 | 2024-07 |
| MiniCPM-V 2.6 8B | OpenBMB | Dense | 8.1B | 32K | 48 | 2024-08 |
| Apertus v1.5 8B | Swiss AI | Dense | 8.9B | 256K | 46 | 2026-07 |
| CodeGeeX4 All 9B | Zhipu AI | Dense | 9.4B | 128K | 46 | 2024-07 |
| DeepHermes 3 Llama 3 8B Preview | Nous Research | Dense | 8.0B | 128K | 46 | 2025-02 |
| DeepSeek R1 Distill Qwen 7B | DeepSeek | Dense | 7.6B | 128K | 46 | 2025-01 |
| Dolphin 3.0 Llama 3.1 8B | Cognitive Computations | Dense | 8.0B | 128K | 46 | 2024-12 |
| EXAONE 3.5 7.8B | LG AI Research | Dense | 7.8B | 32K | 46 | 2024-12 |
| Falcon 3 7B Instruct | TII | Dense | 7.5B | 32K | 46 | 2024-12 |
| Gemma 3n E4B | Dense | 7.9B | 32K | 46 | 2025-06 | |
| Granite 3.2 8B Instruct | IBM | Dense | 8.2B | 128K | 46 | 2025-02 |
| Granite 3.3 8B | IBM | Dense | 8.2B | 128K | 46 | 2025-04 |
| Granite 4.1 3B | IBM | Dense | 3.4B | 128K | 46 | 2026-04 |
| Ministral 8B (2410) | Mistral AI | Dense | 8.0B | 32K | 46 | 2024-10 |
| Moonlight 16B-A3B Instruct | Moonshot AI | Mixture of experts | 16B (3.0B active) | 8K | 46 | 2025-02 |
| Phi-3 Medium (14B) | Microsoft | Dense | 14B | 4K | 46 | 2024-05 |
| Phi-3 Medium 128K Instruct | Microsoft | Dense | 14B | 128K | 46 | 2024-05 |
| Phi-4 Mini Reasoning (3.8B) | Microsoft | Dense | 3.8B | 128K | 46 | 2025-04 |
| Qwen2.5-Omni 7B | Alibaba | Dense | 11B | 32K | 46 | 2025-03 |
| Tülu 3 8B | Ai2 | Dense | 8.0B | 128K | 46 | 2024-11 |
| Gemma 3 4B IT | Dense | 4.3B | 128K | 45 | 2025-03 | |
| Aya Vision 8B | Cohere | Dense | 8.6B | 16K | 44 | 2025-03 |
| Code Llama 70B Instruct | Meta | Dense | 69B | 4K | 44 | 2024-01 |
| Command R7B | Cohere | Dense | 8.0B | 128K | 44 | 2024-12 |
| DeepSeek Coder 33B Instruct | DeepSeek | Dense | 33B | 16K | 44 | 2023-11 |
| Dolphin 2.9.4 Llama 3.1 8B | Cognitive Computations | Dense | 8.0B | 128K | 44 | 2024-08 |
| Granite 3.1 8B Instruct | IBM | Dense | 8.2B | 128K | 44 | 2024-12 |
| Granite 4.0 H Tiny (7B-A1B) | IBM | Mixture of experts | 6.9B (1.0B active) | 128K | 44 | 2025-09 |
| Hermes 2 Pro Llama 3 8B | Nous Research | Dense | 8.0B | 8K | 44 | 2024-04 |
| InternVL2 8B | Shanghai AI Lab | Dense | 8.1B | 32K | 44 | 2024-06 |
| Jamba Reasoning 3B | AI21 Labs | Dense | 3.2B | 256K | 44 | 2025-10 |
| LFM2.5 2.6B | Liquid AI | Dense | 2.7B | 128K | 44 | 2026-07 |
| LFM2.5-VL 3B | Liquid AI | Dense | 3.1B | 32K | 44 | 2026-08 |
| Llama 3.1 Nemotron Nano 4B v1.1 | NVIDIA | Dense | 4.5B | 128K | 44 | 2025-05 |
| LLaVA 1.6 34B | LLaVA | Dense | 35B | 4K | 44 | 2024-03 |
| MedGemma 4B IT | Dense | 4.3B | 128K | 44 | 2025-05 | |
| Ministral 3 3B | Mistral AI | Dense | 3.9B | 256K | 44 | 2025-12 |
| North Micro Vision | Cohere | Dense | 2.5B | 488K | 44 | 2026-08 |
| OLMo 2 13B Instruct | Ai2 | Dense | 14B | 4K | 44 | 2024-12 |
| OpenChat 3.6 8B | OpenChat | Dense | 8.0B | 8K | 44 | 2024-05 |
| Phi-3 Small 8K Instruct | Microsoft | Dense | 7.4B | 8K | 44 | 2024-05 |
| Phi-4 Mini (3.8B) | Microsoft | Dense | 3.8B | 128K | 44 | 2025-02 |
| Phi-4 Multimodal Instruct | Microsoft | Dense | 5.6B | 128K | 44 | 2025-02 |
| Qwen1.5 32B Chat | Alibaba | Dense | 33B | 32K | 44 | 2024-04 |
| Qwen2 7B Instruct | Alibaba | Dense | 7.6B | 32K | 44 | 2024-06 |
| Qwen2-VL 7B Instruct | Alibaba | Dense | 8.3B | 32K | 44 | 2024-08 |
| Command R7B Arabic | Cohere | Dense | 8.0B | 128K | 42 | 2025-02 |
| Hunyuan 4B Instruct | Tencent | Dense | 4.2B | 256K | 42 | 2025-07 |
| InternLM2 Chat 20B | Shanghai AI Lab | Dense | 20B | 32K | 42 | 2024-01 |
| Jamba2 3B | AI21 Labs | Dense | 3.0B | 256K | 42 | 2026-01 |
| Llama 3 8B Instruct | Meta | Dense | 8.0B | 8K | 42 | 2024-04 |
| Llama 3.2 11B Vision Instruct | Meta | Dense | 11B | 128K | 42 | 2024-09 |
| Mistral-NeMo-Minitron 8B Instruct | NVIDIA | Dense | 8.4B | 8K | 42 | 2024-10 |
| SmolLM3 3B | Hugging Face | Dense | 3.1B | 64K | 42 | 2025-07 |
| AFM 4.5B | Arcee AI | Dense | 4.6B | 64K | 40 | 2025-07 |
| Apertus 8B Instruct | Swiss AI | Dense | 8.1B | 64K | 40 | 2025-09 |
| Aya Expanse 8B | Cohere | Dense | 8.0B | 8K | 40 | 2024-10 |
| DeepSeek LLM 67B Chat | DeepSeek | Dense | 67B | 4K | 40 | 2023-11 |
| DeepSeek Prover V2 7B | DeepSeek | Dense | 6.9B | 64K | 40 | 2025-04 |
| DeepSeek V2 Lite Chat | DeepSeek | Mixture of experts | 16B (2.4B active) | 160K | 40 | 2024-05 |
| Granite 3.0 8B Instruct | IBM | Dense | 8.2B | 4K | 40 | 2024-10 |
| Granite 4.0 Micro (3B) | IBM | Dense | 3.4B | 128K | 40 | 2025-09 |
| Granite Code 34B Instruct | IBM | Dense | 34B | 8K | 40 | 2024-04 |
| InternLM2 Math Plus 20B | Shanghai AI Lab | Dense | 20B | 8K | 40 | 2024-05 |
| InternLM2.5 7B Chat | Shanghai AI Lab | Dense | 7.7B | 32K | 40 | 2024-06 |
| Nous Hermes 2 SOLAR 10.7B | Nous Research | Dense | 11B | 4K | 40 | 2024-01 |
| Phi-3.5 Mini Instruct | Microsoft | Dense | 3.8B | 128K | 40 | 2024-08 |
| Qwen3.5 2B | Alibaba | Dense | 2.3B | 256K | 40 | 2026-02 |
| SOLAR 10.7B Instruct | Upstage | Dense | 11B | 4K | 40 | 2023-12 |
| StarCoder2 15B Instruct | BigCode | Dense | 16B | 16K | 40 | 2024-04 |
| TMax 2B | Ai2 | Dense | 1.9B | 256K | 40 | 2026-06 |
| Unlimited OCR (3B-A0.6B) | Baidu | Mixture of experts | 3.3B (0.6B active) | 32K | 40 | 2026-06 |
| Yi 34B Chat | 01.AI | Dense | 34B | 4K | 40 | 2023-11 |
| Yi Coder 9B Chat | 01.AI | Dense | 8.8B | 128K | 40 | 2024-08 |
| Code Llama 34B Instruct | Meta | Dense | 34B | 16K | 38 | 2023-08 |
| CogVLM2 Llama 3 Chat 19B | Zhipu AI | Dense | 20B | 8K | 38 | 2024-05 |
| Dolphin 2.9 Llama 3 8B | Cognitive Computations | Dense | 8.0B | 8K | 38 | 2024-04 |
| EXAONE 3.0 7.8B Instruct | LG AI Research | Dense | 7.8B | 4K | 38 | 2024-07 |
| Falcon3 Mamba 7B Instruct | TII | Dense | 7.3B | 32K | 38 | 2024-12 |
| Granite 4.0 H Micro | IBM | Dense | 3.2B | 128K | 38 | 2025-09 |
| LFM2 8B-A1B | Liquid AI | Mixture of experts | 8.3B (1.5B active) | 125K | 38 | 2025-10 |
| MiniCPM3 4B | OpenBMB | Dense | 4.1B | 32K | 38 | 2024-09 |
| Mistral 7B Instruct v0.3 | Mistral AI | Dense | 7.3B | 32K | 38 | 2024-05 |
| Phi-3.5 Vision Instruct | Microsoft | Dense | 4.2B | 128K | 38 | 2024-08 |
| Zamba2 7B Instruct | Zyphra | Dense | 7.4B | 4K | 38 | 2024-10 |
| Codestral Mamba 7B | Mistral AI | Dense | 7.3B | 256K | 36 | 2024-07 |
| Falcon-H1 3B Instruct | TII | Dense | 3.2B | 128K | 36 | 2025-05 |
| Gemma 3n E2B | Dense | 5.4B | 32K | 36 | 2025-06 | |
| Granite Code 20B Instruct | IBM | Dense | 20B | 8K | 36 | 2024-04 |
| Hermes 2 Pro Mistral 7B | Nous Research | Dense | 7.2B | 32K | 36 | 2024-03 |
| Llama 2 70B Chat | Meta | Dense | 69B | 4K | 36 | 2023-07 |
| OLMo 2 7B Instruct | Ai2 | Dense | 7.3B | 4K | 36 | 2024-12 |
| Phi-3 Mini 128K Instruct | Microsoft | Dense | 3.8B | 128K | 36 | 2024-04 |
| Phi-3 Mini 4K Instruct | Microsoft | Dense | 3.8B | 4K | 36 | 2024-04 |
| Qwen2.5 Coder 3B Instruct | Alibaba | Dense | 3.1B | 32K | 36 | 2024-11 |
| Qwen2.5-Math 7B Instruct | Alibaba | Dense | 7.6B | 4K | 36 | 2024-09 |
| Qwen3-VL 2B Instruct | Alibaba | Dense | 2.1B | 256K | 36 | 2025-10 |
| Starling LM 7B Beta | Nexusflow | Dense | 7.2B | 8K | 36 | 2024-03 |
| Voxtral Mini 3B | Mistral AI | Dense | 4.7B | 32K | 36 | 2025-07 |
| Yi 1.5 9B Chat | 01.AI | Dense | 8.8B | 4K | 36 | 2024-05 |
| Llama 3.2 3B Instruct | Meta | Dense | 3.2B | 128K | 35 | 2024-09 |
| OpenChat 3.5 (0106) | OpenChat | Dense | 7.2B | 8K | 35 | 2024-01 |
| Aya 23 8B | Cohere | Dense | 8.0B | 8K | 34 | 2024-05 |
| CodeGemma 7B | Dense | 8.5B | 8K | 34 | 2024-04 | |
| DeepSeek Coder 6.7B Instruct | DeepSeek | Dense | 6.7B | 16K | 34 | 2023-10 |
| Falcon 180B Chat | TII | Dense | 180B | 2K | 34 | 2023-09 |
| LFM2 2.6B | Liquid AI | Dense | 2.6B | 125K | 34 | 2025-09 |
| Mathstral 7B | Mistral AI | Dense | 7.3B | 32K | 34 | 2024-07 |
| Mistral 7B Instruct v0.2 | Mistral AI | Dense | 7.2B | 32K | 34 | 2023-12 |
| Nous Hermes 2 Mistral 7B DPO | Nous Research | Dense | 7.2B | 32K | 34 | 2024-02 |
| OpenChat 3.5 (1210) | OpenChat | Dense | 7.2B | 8K | 34 | 2023-12 |
| Phi-3 Vision 128K Instruct | Microsoft | Dense | 4.2B | 128K | 34 | 2024-05 |
| Qwen1.5 14B Chat | Alibaba | Dense | 14B | 32K | 34 | 2024-01 |
| Qwen2.5 3B Instruct | Alibaba | Dense | 3.1B | 32K | 34 | 2024-09 |
| Qwen2.5-VL 3B Instruct | Alibaba | Dense | 3.8B | 125K | 34 | 2025-01 |
| RecurrentGemma 9B IT | Dense | 9.6B | 8K | 34 | 2024-06 | |
| Granite Code 8B Instruct (128K) | IBM | Dense | 8.1B | 125K | 32 | 2024-07 |
| Hermes 3 Llama 3.2 3B | Nous Research | Dense | 3.2B | 128K | 32 | 2024-12 |
| InternLM2 Math Plus 7B | Shanghai AI Lab | Dense | 7.7B | 8K | 32 | 2024-05 |
| LFM2.5 1.2B Instruct | Liquid AI | Dense | 1.2B | 125K | 32 | 2026-01 |
| OpenHermes 2.5 Mistral 7B | Teknium | Dense | 7.2B | 32K | 32 | 2023-10 |
| StarCoder2 15B (base) | BigCode | Dense | 16B | 16K | 32 | 2024-02 |
| Starling LM 7B Alpha | Berkeley | Dense | 7.2B | 8K | 32 | 2023-11 |
| Code Llama 13B Instruct | Meta | Dense | 13B | 16K | 30 | 2023-08 |
| EXAONE 3.5 2.4B | LG AI Research | Dense | 2.4B | 32K | 30 | 2024-12 |
| EXAONE 4.0 1.2B | LG AI Research | Dense | 1.3B | 64K | 30 | 2025-07 |
| Falcon 3 3B Instruct | TII | Dense | 3.2B | 32K | 30 | 2024-12 |
| Falcon Mamba 7B Instruct | TII | Dense | 7.3B | 128K | 30 | 2024-07 |
| Granite 3.2 2B Instruct | IBM | Dense | 2.5B | 128K | 30 | 2025-02 |
| Granite 3.3 2B | IBM | Dense | 2.5B | 128K | 30 | 2025-04 |
| Mistral 7B Instruct v0.1 | Mistral AI | Dense | 7.2B | 8K | 30 | 2023-09 |
| Qwen2.5-Omni 3B | Alibaba | Dense | 5.5B | 32K | 30 | 2025-04 |
| Qwen3 1.7B | Alibaba | Dense | 2.0B | 40K | 30 | 2025-04 |
| SQLCoder 7B v2 | Defog | Dense | 6.7B | 16K | 30 | 2024-02 |
| Gemma 1.1 7B IT | Dense | 8.5B | 8K | 28 | 2024-03 | |
| Gemma 2 2B | Dense | 2.6B | 8K | 28 | 2024-07 | |
| Granite 3.1 2B Instruct | IBM | Dense | 2.5B | 128K | 28 | 2024-12 |
| Granite 4.0 H 1B | IBM | Dense | 1.5B | 128K | 28 | 2025-10 |
| LLaVA 1.6 Mistral 7B | LLaVA | Dense | 7.6B | 32K | 28 | 2024-02 |
| Neural Chat 7B v3.3 | Intel | Dense | 7.2B | 32K | 28 | 2023-12 |
| Orca 2 13B | Microsoft | Dense | 13B | 4K | 28 | 2023-11 |
| StableLM 2 12B Chat | Stability AI | Dense | 12B | 4K | 28 | 2024-04 |
| Yi 1.5 6B Chat | 01.AI | Dense | 6.1B | 4K | 28 | 2024-05 |
| Baichuan 2 13B Chat | Baichuan | Dense | 14B | 4K | 26 | 2023-08 |
| DeepScaleR 1.5B Preview | Agentica | Dense | 1.8B | 128K | 26 | 2025-01 |
| DeepSeek MoE 16B Chat | DeepSeek | Mixture of experts | 16B (2.8B active) | 4K | 26 | 2024-01 |
| Falcon-H1 1.5B Instruct | TII | Dense | 1.6B | 128K | 26 | 2025-05 |
| Gemma 7B IT | Dense | 8.5B | 8K | 26 | 2024-02 | |
| Granite 3.0 2B Instruct | IBM | Dense | 2.6B | 4K | 26 | 2024-10 |
| Granite Vision 3.2 2B | IBM | Dense | 3.0B | 128K | 26 | 2025-02 |
| LFM2 1.2B | Liquid AI | Dense | 1.2B | 125K | 26 | 2025-07 |
| Llama 2 13B Chat | Meta | Dense | 13B | 4K | 26 | 2023-07 |
| OLMoE 1B-7B Instruct | Ai2 | Mixture of experts | 6.9B (1.3B active) | 4K | 26 | 2025-01 |
| Qwen1.5 7B Chat | Alibaba | Dense | 7.7B | 32K | 26 | 2024-01 |
| Qwen2.5 Coder 1.5B Instruct | Alibaba | Dense | 1.5B | 32K | 26 | 2024-09 |
| Qwen3.5 0.8B | Alibaba | Dense | 0.9B | 256K | 26 | 2026-02 |
| StarCoder 15B | BigCode | Dense | 16B | 8K | 26 | 2023-04 |
| WizardMath 7B v1.1 | WizardLM | Dense | 7.2B | 32K | 26 | 2023-12 |
| Zephyr 7B Beta | Hugging Face | Dense | 7.2B | 32K | 26 | 2023-10 |
| Code Llama 7B Instruct | Meta | Dense | 6.7B | 16K | 24 | 2023-08 |
| LFM2-VL 1.6B | Liquid AI | Dense | 1.6B | 32K | 24 | 2025-08 |
| Qwen1.5 MoE A2.7B Chat | Alibaba | Mixture of experts | 14B (2.7B active) | 8K | 24 | 2024-03 |
| Qwen2-VL 2B Instruct | Alibaba | Dense | 2.2B | 32K | 24 | 2024-08 |
| Qwen2.5 1.5B Instruct | Alibaba | Dense | 1.5B | 32K | 24 | 2024-09 |
| SmolVLM2 2.2B Instruct | Hugging Face | Dense | 2.3B | 8K | 24 | 2025-02 |
| StarCoder2 7B | BigCode | Dense | 7.2B | 16K | 24 | 2024-02 |
| DeepSeek R1 Distill Qwen 1.5B | DeepSeek | Dense | 1.8B | 128K | 22 | 2025-01 |
| Hunyuan 1.8B Instruct | Tencent | Dense | 1.8B | 256K | 22 | 2025-07 |
| Llama 3.2 1B Instruct | Meta | Dense | 1.2B | 128K | 22 | 2024-09 |
| LLaVA 1.5 13B | LLaVA | Dense | 13B | 4K | 22 | 2023-10 |
| Nemotron Mini 4B Instruct | NVIDIA | Dense | 4.2B | 4K | 22 | 2024-09 |
| Orca 2 7B | Microsoft | Dense | 6.7B | 4K | 22 | 2023-11 |
| SmolVLM 2.2B Instruct | Hugging Face | Dense | 2.3B | 16K | 22 | 2024-11 |
| WizardLM 13B v1.2 | WizardLM | Dense | 13B | 4K | 22 | 2023-07 |
| Zamba2 2.7B Instruct | Zyphra | Dense | 2.7B | 4K | 22 | 2024-08 |
| Baichuan 2 7B Chat | Baichuan | Dense | 7.5B | 4K | 20 | 2023-08 |
| Gemma 3 1B | Dense | 1.0B | 32K | 20 | 2025-03 | |
| Granite 3.1 3B-A800M Instruct | IBM | Mixture of experts | 3.3B (0.8B active) | 128K | 20 | 2024-12 |
| Granite Code 3B Instruct (128K) | IBM | Dense | 3.5B | 125K | 20 | 2024-07 |
| Llama 2 7B Chat | Meta | Dense | 6.7B | 4K | 20 | 2023-07 |
| Qwen2.5-Math 1.5B Instruct | Alibaba | Dense | 1.5B | 4K | 20 | 2024-09 |
| SmolLM2 1.7B Instruct | Hugging Face | Dense | 1.7B | 8K | 20 | 2024-10 |
| Stable Code Instruct 3B | Stability AI | Dense | 2.8B | 16K | 20 | 2024-03 |
| Vicuna 13B v1.5 | LMSYS | Dense | 13B | 4K | 20 | 2023-07 |
| Yi 6B Chat | 01.AI | Dense | 6.1B | 4K | 20 | 2023-11 |
| DeepSeek LLM 7B Chat | DeepSeek | Dense | 6.9B | 4K | 18 | 2023-11 |
| Granite 3.0 3B-A800M Instruct | IBM | Mixture of experts | 3.4B (0.8B active) | 4K | 18 | 2024-10 |
| InternLM2.5 1.8B Chat | Shanghai AI Lab | Dense | 1.9B | 32K | 18 | 2024-07 |
| LFM2 700M | Liquid AI | Dense | 0.7B | 125K | 18 | 2025-07 |
| StarCoder2 3B | BigCode | Dense | 3.0B | 16K | 18 | 2024-02 |
| Falcon 3 1B Instruct | TII | Dense | 1.7B | 8K | 16 | 2024-12 |
| Falcon 40B Instruct | TII | Dense | 42B | 2K | 16 | 2023-05 |
| LLaVA 1.5 7B | LLaVA | Dense | 7.1B | 4K | 16 | 2023-10 |
| Phi-2 2.7B | Microsoft | Dense | 2.8B | 2K | 16 | 2023-12 |
| Qwen2 1.5B Instruct | Alibaba | Dense | 1.5B | 32K | 16 | 2024-06 |
| StableLM Zephyr 3B | Stability AI | Dense | 2.8B | 4K | 16 | 2023-11 |
| Yi Coder 1.5B Chat | 01.AI | Dense | 1.5B | 128K | 16 | 2024-08 |
| CodeGemma 2B | Dense | 2.5B | 8K | 14 | 2024-03 | |
| Falcon-H1 0.5B Instruct | TII | Dense | 0.5B | 16K | 14 | 2025-05 |
| LFM2.5 350M | Liquid AI | Dense | 0.4B | 125K | 14 | 2026-03 |
| OLMo 2 1B Instruct | Ai2 | Dense | 1.5B | 4K | 14 | 2025-04 |
| OLMo 7B Instruct | Ai2 | Dense | 6.9B | 2K | 14 | 2024-06 |
| Qwen3 0.6B | Alibaba | Dense | 0.8B | 40K | 14 | 2025-04 |
| SmolLM 1.7B Instruct | Hugging Face | Dense | 1.7B | 2K | 14 | 2024-07 |
| Vicuna 7B v1.5 | LMSYS | Dense | 6.7B | 4K | 14 | 2023-07 |
| DeepSeek Coder 1.3B Instruct | DeepSeek | Dense | 1.4B | 16K | 12 | 2023-10 |
| Gemma 2B IT | Dense | 2.5B | 8K | 12 | 2024-02 | |
| Granite 4.0 H 350M | IBM | Dense | 0.3B | 32K | 12 | 2025-10 |
| InternLM 7B | Shanghai AI Lab | Dense | 7.3B | 2K | 12 | 2023-07 |
| LFM2 350M | Liquid AI | Dense | 0.4B | 125K | 12 | 2025-07 |
| LFM2-VL 450M | Liquid AI | Dense | 0.5B | 32K | 12 | 2025-08 |
| LFM2.5 230M | Liquid AI | Dense | 0.2B | 125K | 10 | 2026-06 |
| Qwen2.5 0.5B Instruct | Alibaba | Dense | 0.5B | 32K | 10 | 2024-09 |
| SmolVLM 500M Instruct | Hugging Face | Dense | 0.5B | 8K | 10 | 2025-01 |
| StableLM 2 1.6B Chat | Stability AI | Dense | 1.6B | 4K | 10 | 2024-04 |
| StableLM 2 Zephyr 1.6B | Stability AI | Dense | 1.6B | 4K | 10 | 2024-01 |
| Hunyuan 0.5B Instruct | Tencent | Dense | 0.5B | 256K | 8 | 2025-07 |
| Qwen2.5 Coder 0.5B Instruct | Alibaba | Dense | 0.5B | 32K | 8 | 2024-11 |
| SmolLM2 360M Instruct | Hugging Face | Dense | 0.4B | 8K | 8 | 2024-10 |
| ERNIE 4.5 0.3B | Baidu | Dense | 0.4B | 128K | 6 | 2025-06 |
| Falcon 7B Instruct | TII | Dense | 7.2B | 2K | 6 | 2023-04 |
| Gemma 3 270M | Dense | 0.3B | 32K | 6 | 2025-08 | |
| Qwen2 0.5B Instruct | Alibaba | Dense | 0.5B | 32K | 6 | 2024-06 |
| SmolVLM 256M Instruct | Hugging Face | Dense | 0.3B | 8K | 6 | 2025-01 |
| TinyLlama 1.1B Chat | TinyLlama | Dense | 1.1B | 2K | 6 | 2023-12 |
| SmolLM 360M Instruct | Hugging Face | Dense | 0.4B | 2K | 5 | 2024-07 |
| SmolLM2 135M Instruct | Hugging Face | Dense | 0.1B | 8K | 4 | 2024-10 |
| Falcon-H1 Tiny 90M Instruct | TII | Dense | 0.1B | 256K | 3 | 2026-01 |
| SmolLM 135M Instruct | Hugging Face | Dense | 0.1B | 2K | 2 | 2024-07 |
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.