Hardware · Clusters · Example cluster

Four Mac minis (M4, 24 GB): what it runs together

Four entry-level 24 GB Mac minis on Thunderbolt 4 cables.

Four entry-level Mac minis hold about 64 GB together. With EXO's tensor split over Thunderbolt they read in parallel, so 27B to 32B models type about twice as fast as on one of them; 70B models fit, but slowly.

Change the devices How clustering works

64.3 GBusable memory together (the largest device has 16.1 GB)

63models only the cluster can run

404run better or faster than on any one device

128Bbiggest model it adds: Mistral Medium 3.5 (128B)

Connected by Thunderbolt 4 or USB4 cable. 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 453 of the 467 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

518 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
87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Bolt 74B-A9BNVIDIA · 75B (9.0B active) Q4_K_M 53.3 GB 35.5 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B)Aleph Alpha · 78B (3.5B active) Q4_K_M 55.4 GB 53.4 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
81Great quality (A) · 81 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. Qwen3 Coder Next (80B-A3B)Alibaba · 80B (3.0B active) Q4_K_M 56.3 GB 51.5 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
81Great quality (A) · 81 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. Qwen3-Next 80B-A3B ThinkingAlibaba · 81B (3.0B active) Q4_K_M 57.4 GB 51.5 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
79Great quality (A) · 79 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Qwen3 Next 80B-A3B InstructAlibaba · 81B (3.0B active) Q4_K_M 57.4 GB 51.5 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
73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. Kimi Linear 48B-A3BMoonshot AI · 49B (3.0B active) Q6_K 47.9 GB 53.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. Ling 2.6 Flash (104B-A7.4B)Ant Group (inclusionAI) · 107B (7.4B active) Q3_K_M 60.3 GB 54.9 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
79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. Llama 3.3 Nemotron Super 49B v1.5NVIDIA · 50B Q6_K 51.1 GB 5.6 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
70Great quality (A) · 70 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. Hunyuan A13B (80B-A13B)Tencent · 80B (13.0B active) Q4_K_M 57.7 GB 18.1 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
65Good quality (B) · 65 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. Ring Flash 2.0 (100B-A6B)Ant Group (inclusionAI) · 103B (6.1B active) Q3_K_M 58.5 GB 26.5 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
69Good quality (B) · 69 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-4.6VZhipu AI · 108B (12.0B active) Q3_K_M 61.9 GB 17.4 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
63Good quality (B) · 63 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. Ling Flash 2.0 (100B-A6B)Ant Group (inclusionAI) · 103B (6.1B active) Q3_K_M 58.5 GB 26.5 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
76Great quality (A) · 76 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. Kimi Dev 72BMoonshot AI · 73B Q4_K_M 54.6 GB 5.6 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
67Good quality (B) · 67 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-4.5VZhipu AI · 108B (12.0B active) Q3_K_M 61.9 GB 17.4 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
67Good quality (B) · 67 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. Solar Open 100BUpstage · 103B (12.0B active) Q3_K_M 59.5 GB 16.9 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
63Good quality (B) · 63 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. Sarvam 105BSarvam AI · 106B (10.0B active) Q3_K_M 59.8 GB 22.8 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
65Good quality (B) · 65 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. GLM-4.5 Air (106B-A12B)Zhipu AI · 110B (12.0B active) Q3_K_M 63.2 GB 17.4 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
70Great quality (A) · 70 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. Hermes 4 70BNous Research · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
70Great quality (A) · 70 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. Llama 3.3 70B InstructMeta · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. Ornith 1.5 122B-A10BDeepReinforce · 125B (10.0B active) Q2_K 57.8 GB 14.4 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.5 122B-A10BAlibaba · 125B (10.0B active) Q2_K 57.8 GB 14.4 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Super VL 120B-A12BNVIDIA · 124B (12.0B active) Q2_K 57.4 GB 12.7 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
69Good quality (B) · 69 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. Qwen2.5 72B InstructAlibaba · 73B Q4_K_M 54.6 GB 5.6 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
68Good quality (B) · 68 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. DeepSeek R1 Distill Llama 70BDeepSeek · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
68Good quality (B) · 68 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. Tülu 3 70BAi2 · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
70Great quality (A) · 70 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. InternVL3 78BShanghai AI Lab · 78B Q4_K_M 58.1 GB 5.3 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
57Good quality (B) · 57 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. Ling 3.0 Flash (127B-A5B)Ant Group (inclusionAI) · 127B (5.1B active) Q2_K 58.9 GB 21.9 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
60Good quality (B) · 60 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. Laguna S 2.1 (118B-A8B)Poolside · 118B (8.0B active) Q2_K 56.2 GB 16.8 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
69Good quality (B) · 69 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. Qwen2.5-VL 72B InstructAlibaba · 73B Q4_K_M 55.0 GB 5.6 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
66Good quality (B) · 66 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. Llama 3.1 70B InstructMeta · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
68Good quality (B) · 68 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. Command A Reasoning (111B)Cohere · 111B Q3_K_M 62.7 GB 5.6 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
60Good quality (B) · 60 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. Llama 4 Scout 17B-16EMeta · 109B (17.0B active) Q3_K_M 62.4 GB 14.9 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
61Good quality (B) · 61 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. Nemotron 3 Super 120B-A12BNVIDIA · 124B (12.0B active) Q2_K 57.1 GB 12.7 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
56Good quality (B) · 56 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 Small 4 (119B-A6B)Mistral AI · 119B (6.0B active) Q2_K 55.5 GB 20.0 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. Command A (111B)Cohere · 111B Q3_K_M 62.7 GB 5.6 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
64Good quality (B) · 64 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. Hermes 3 (Llama 3.1 70B)Nous Research · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
47Fair quality (C) · 47 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. Jamba 1.5 MiniAI21 Labs · 52B (12.0B active) Q4_K_M 39.1 GB 32.4 tok/s Runs great Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
62Good quality (B) · 62 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. Llama 3.1 Nemotron 70B InstructNVIDIA · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
62Good quality (B) · 62 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. Apertus v1.5 70BSwiss AI · 72B Q4_K_M 54.2 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
64Good quality (B) · 64 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. QVQ 72B PreviewAlibaba · 73B Q4_K_M 55.0 GB 5.6 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
60Good quality (B) · 60 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. Qwen2 72B InstructAlibaba · 73B Q4_K_M 54.6 GB 5.6 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
60Good quality (B) · 60 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. Qwen2-VL 72B InstructAlibaba · 73B Q4_K_M 55.0 GB 5.6 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. Groot 65B A6B 261023GrootTeam · 118B (6.0B active) Q2_K 55.1 GB 20.0 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
47Fair quality (C) · 47 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. Qwen2 57B-A14B InstructAlibaba · 57B (14.0B active) Q4_K_M 43.0 GB 18.5 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. Ling 3.0 flash SanteAnt Group (inclusionAI) · 127B (5.1B active) Q2_K 58.6 GB 21.9 tok/s Runs welltight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Pipeline split llama.cpp RPC Needs the clusterWon’t run on any one of them
57Good quality (B) · 57 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. Apertus 70B InstructSwiss AI · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. Command R+ (08-2024) 104BCohere · 104B Q3_K_M 60.6 GB 5.4 tok/s Slowtight fit Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
49Fair quality (C) · 49 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. Nous Hermes 2 Mixtral 8x7B DPONous Research · 47B (12.9B active) Q4_K_M 37.1 GB 18.1 tok/s Runs well Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. Llama 3.2 90B Vision InstructMeta · 89B Q3_K_M 54.3 GB 5.6 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them
55Good quality (B) · 55 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. Llama 3 70B InstructMeta · 71B Q4_K_M 53.3 GB 5.7 tok/s Slow Mac 1 + Mac 2 + Mac 3 + Mac 4Tensor split EXO Needs the clusterWon’t run on any one of them

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?