Hardware · Clusters · Example cluster

Four Mac Studios (M3 Ultra, 512 GB) with RDMA: what it runs together

Four 512 GB Mac Studios, every pair joined by Thunderbolt 5 with RDMA (macOS 26.2 or later).

The setup behind EXO's frontier-model demos: about 1.5 TB of usable memory runs trillion-parameter models such as Kimi K2 at 4 to 6 bits, and the tensor split over RDMA types faster than one Mac Studio. Kimi K3 (2.8 trillion parameters, native 4-bit) only fits after a further cut to 3-bit; the one published 4-bit run on four Mac Studios typed about 2 tokens per second, far below our estimate, so treat that row as a best case.

Change the devices How clustering works

1,536 GBusable memory together (the largest device has 384 GB)

5models only the cluster can run

538run better or faster than on any one device

2,780B (104.0B active)biggest model it adds: Kimi K3.1 Pro (2.8T)

Connected by Thunderbolt 5 with RDMA (macOS 26.2+). The wait for the first word is worked out for a 1,000-token prompt (a short chat).

The devices

Tool to use

EXO best estimate for 543 of the 543 models the cluster adds

Two or more Apple silicon Macs, best of all Mac Studios or Mac minis joined by Thunderbolt 5 cables, when you want big models with almost no setup.

brew install --cask exo   # on every Mac, then open EXO
open http://localhost:52415   # pick a model and how to split it

619 LLMs this cluster can run

Models the cluster adds come first, best match first, then the ones a single device already runs as well. Filter vs one device to pick. Speeds are estimates for one person chatting.

Quality Model Quant Memory Speed Runs Devices Tool vs one device
91Excellent quality (A+) · 91 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V4 Pro (1.6T-A49B)DeepSeek · 1,651B (49.0B active) Q4_K_M 1,133 GB 33.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V4 Pro (April 2026)DeepSeek · 1,599B (49.0B active) Q4_K_M 1,101 GB 33.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
82Great quality (A) · 82 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.8 Max (2.4T-A95B)Alibaba · 2,446B (95.0B active) Q3_K_M 1,325 GB 34.1 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
83Great quality (A) · 83 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Kimi K3.1 Pro (2.8T)Moonshot AI · 2,780B (104.0B active) Q3_K_M 1,488 GB 31.7 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
82Great quality (A) · 82 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Kimi K3 (2.8T)Moonshot AI · 2,780B (104.0B active) Q3_K_M 1,488 GB 31.9 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
98Excellent quality (A+) · 98 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V4.1 Flash 1017 (552B-A16B)DeepSeek · 763B (16.0B active) Q8_0 934 GB 69.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 69.5 tok/s
96Excellent quality (A+) · 96 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.6 Pro (1T-A42B)Xiaomi · 1,024B (42.0B active) Q6_K 966 GB 41.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 26.0 tok/s
96Excellent quality (A+) · 96 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V4.1 Flash (552B-A16B)DeepSeek · 763B (16.0B active) Q8_0 934 GB 69.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 69.5 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.5 Pro (1T-A42B)Xiaomi · 1,023B (42.0B active) Q6_K 965 GB 41.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 26.0 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Kimi K2.6 (1T)Moonshot AI · 1,027B (32.0B active) Q6_K 968 GB 33.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 33.2 tok/s
93Excellent quality (A+) · 93 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Kimi K2.5 (1T)Moonshot AI · 1,027B (32.0B active) Q6_K 968 GB 33.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 33.2 tok/s
93Excellent quality (A+) · 93 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Kimi K2.7 Code (1T)Moonshot AI · 1,027B (32.0B active) Q6_K 968 GB 33.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 33.2 tok/s
92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Inkling (975B-A41B)Thinking Machines Lab · 952B (41.0B active) Q6_K 907 GB 42.5 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 26.6 tok/s
97Excellent quality (A+) · 97 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Step 5 Preview (600B-A27B)StepFun · 604B (27.0B active) Q8_0 766 GB 40.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 36.5 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. Mistral Large 4 (1T-A52B)Mistral AI · 1,053B (52.0B active) Q6_K 990 GB 25.9 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 21.4 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. Muse Spark Open (816B-A42B)Meta · 822B (42.0B active) Q6_K 800 GB 40.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 30.6 tok/s
97Excellent quality (A+) · 97 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. Ling 3.1 Flash 1019 (560B-A25B)Ant Group (inclusionAI) · 572B (25.0B active) Q8_0 732 GB 48.7 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 31.0 tok/s
92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. Kimi K2 Thinking (1T)Moonshot AI · 1,026B (32.0B active) Q6_K 968 GB 33.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 33.2 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. Ling 3.1 Flash (560B-A25B)Ant Group (inclusionAI) · 572B (25.0B active) Q8_0 732 GB 48.7 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 31.0 tok/s
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. Kimi K2 Instruct 0905 (1T)Moonshot AI · 1,026B (32.0B active) Q6_K 968 GB 33.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 33.2 tok/s
88Excellent quality (A+) · 88 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Kimi K2 Instruct (July 2025)Moonshot AI · 1,026B (32.0B active) Q6_K 968 GB 33.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 33.2 tok/s
97Excellent quality (A+) · 97 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. GLM-5.3 (753B-A40B)Zhipu AI · 753B (40.0B active) Q6_K 744 GB 27.4 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 32.0 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Beam 1.0 (501B-A23B)Reflection AI · 501B (23.4B active) Q8_0 657 GB 51.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 32.9 tok/s
93Excellent quality (A+) · 93 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Inkling Small (276B-A12B)Thinking Machines Lab · 266B (12.0B active) F16 657 GB 45.9 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 42.4 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. GLM-5.2 (753B-A40B)Zhipu AI · 753B (40.0B active) Q6_K 744 GB 27.4 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 32.0 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Intern-S2 (397B-A17B)Shanghai AI Lab · 403B (17.0B active) Q8_0 553 GB 58.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 36.7 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.5 397B-A17BAlibaba · 403B (17.0B active) Q8_0 553 GB 58.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 36.7 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Nex N2 Pro (397B-A17B)Nex-AGI · 397B (17.0B active) Q8_0 546 GB 58.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 36.7 tok/s
89Excellent quality (A+) · 89 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Ling 1TAnt Group (inclusionAI) · 1,000B (50.0B active) Q6_K 948 GB 24.3 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 22.2 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. GLM-5.1 (744B-A40B)Zhipu AI · 754B (40.0B active) Q6_K 744 GB 27.4 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 32.0 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. Hunyuan Hy4 Preview (780B)Tencent · 780B (55.0B active) Q6_K 766 GB 23.3 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 23.9 tok/s
87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. Ling 2.6 1TAnt Group (inclusionAI) · 1,025B (63.0B active) Q4_K_M 751 GB 41.6 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Slow on Mac 1, 17.9 tok/s
93Excellent quality (A+) · 93 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. GLM-5 (744B-A40B)Zhipu AI · 754B (40.0B active) Q6_K 744 GB 27.4 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 32.0 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. Ornith 1.5 397B-A17BDeepReinforce · 403B (17.0B active) Q8_0 553 GB 58.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 36.7 tok/s
96Excellent quality (A+) · 96 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. GLM-5.3 Flash (320B-A18B)Zhipu AI · 321B (18.0B active) Q8_0 466 GB 59.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 41.4 tok/s
92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V3.2 (671B)DeepSeek · 685B (37.0B active) Q6_K 688 GB 31.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 27.5 tok/s
96Excellent quality (A+) · 96 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.6 Flash (310B-A15B)Xiaomi · 311B (15.0B active) Q8_0 455 GB 66.5 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 40.9 tok/s
88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. LongCat Flash Chat (560B)Meituan · 562B (27.0B active) Q8_0 721 GB 38.9 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 28.9 tok/s
96Excellent quality (A+) · 96 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. MiniMax M3.1 Flash (428B-A23B)MiniMax · 429B (23.0B active) Q6_K 478 GB 38.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 33.4 tok/s
92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek R1 0528 (671B)DeepSeek · 685B (37.0B active) Q6_K 687 GB 31.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 27.5 tok/s
93Excellent quality (A+) · 93 of 100: how capable the model is at F16, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. Step 3.7 Flash (198B-A11B)StepFun · 201B (11.0B active) F16 528 GB 46.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 45.6 tok/s
91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V3.2 ExpDeepSeek · 685B (37.0B active) Q6_K 688 GB 31.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 27.5 tok/s
95Excellent quality (A+) · 95 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Naive N0.5 Flash (309B-A15.5B)Naive AI · 309B (15.5B active) Q8_0 453 GB 65.3 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 39.7 tok/s
91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek R1 (671B)DeepSeek · 671B (37.0B active) Q6_K 676 GB 31.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 27.5 tok/s
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V3.1 TerminusDeepSeek · 685B (37.0B active) Q6_K 687 GB 31.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 27.5 tok/s
92Excellent quality (A+) · 92 of 100: how capable the model is at F16, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Step 3.5 Flash (196B-A11B)StepFun · 199B (11.0B active) F16 524 GB 46.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 45.6 tok/s
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V3.1 (671B)DeepSeek · 685B (37.0B active) Q6_K 687 GB 31.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 27.5 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.5 (310B-A15B)Xiaomi · 311B (15.0B active) Q8_0 455 GB 66.5 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 40.9 tok/s
92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3 Ultra 550B-A55BNVIDIA · 561B (55.0B active) Q6_K 585 GB 33.8 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs well on Mac 1, 14.9 tok/s
94Excellent quality (A+) · 94 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. DeepSeek V4 Flash (284B-A13B)DeepSeek · 304B (13.0B active) Q8_0 448 GB 48.0 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Better on the clusterAlone: Runs great on Mac 1, 39.6 tok/s

Other example clusters

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.

Was this page helpful?