GPUs / Apple What AI models can an M2 Ultra (64–192 GB) run? The M2 Ultra has 192 GB of unified memory and 800 GB/s memory bandwidth (speeds below assume 70% of it, the share llama.cpp reaches on dual-die Ultra chips). 90 models run well on it and 2 more are a tight fit. Best models for 128 GB →
Memory 64 GB 128 GB 192 GB
As an Amazon Associate, I earn from qualifying purchases.
Best for general use Highest MMLU score among models that run well
Qwen3 Next 80B-A3B Q8_0 256K context · ~100 tok/s · MMLU 90 Mistral Small 4 Q8_0 135K context · ~60 tok/s · MMLU 88.5 MiniMax M2.1 Q4_K_M 128K context · ~66 tok/s · MMLU 88 Best for coding Highest SWE-bench Verified score among models that run well
Qwen3.8 27B Q8_0 256K context · ~19 tok/s · SWE-bench 86 Ornith 1.5 122B Q8_0 256K context · ~43 tok/s · SWE-bench 84.5 MiniMax M2.5 Q4_K_M 192K context · ~66 tok/s · SWE-bench 80.2 Every model on the M2 Ultra (192 GB) 178 of 214 model configurations can run on 192 GB — sorted by quality
107 models, each counted once by the best way it runs on your hardware.
Largest that runs well GLM 5.3 Flash 321.32B params · Q4_K_M · 1M context Fastest good-quality model Phi-3 Mini 3.8B ~165 tok/s · Q4_K_M · 128K context Runs well 90 View all 90 →
Runs well 90 ✕ Q4_K_M Context 256K Speed ~133 tok/s
MoE — speed based on ~1.69 GB active experts plus per-token routing overhead (full 45.09 GB loaded in VRAM)
Q8_0 Context 256K Speed ~100 tok/s
MoE — speed based on ~2.96 GB active experts plus per-token routing overhead (full 78.99 GB loaded in VRAM)
Q4_K_M Context 226K Speed ~88 tok/s
MoE — speed based on ~3.7 GB active experts plus per-token routing overhead (full 67.65 GB loaded in VRAM)
Q8_0 Context 135K Speed ~60 tok/s
MoE — speed based on ~6.43 GB active experts plus per-token routing overhead (full 117.79 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~66 tok/s
MoE — speed based on ~5.63 GB active experts plus per-token routing overhead (full 128.84 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 226.5 GB
Q4_K_M Context 378K Speed ~54 tok/s
MoE — speed based on ~7.37 GB active experts plus per-token routing overhead (full 161 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 282.1 GB
Q4_K_M Context 256K Speed ~32 tok/s
Q8_0 Context 256K Speed ~18 tok/s
Q4_K_M Context 1M Speed ~43 tok/s
MoE — speed based on ~9.86 GB active experts plus per-token routing overhead (full 176.02 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 310.4 GB
Q4_K_M Context 256K Speed ~32 tok/s
Q8_0 Context 256K Speed ~19 tok/s
Q4_K_M Context 128K Speed ~12 tok/s
Q8_0 Context 128K Speed ~7 tok/s
Q4_K_M Context 128K Speed ~12 tok/s
Q8_0 Context 128K Speed ~7 tok/s
Q4_K_M Context 128K Speed ~13 tok/s
Q8_0 Context 128K Speed ~7 tok/s
Q4_K_M Context 256K Speed ~131 tok/s
MoE — speed based on ~1.76 GB active experts plus per-token routing overhead (full 20.5 GB loaded in VRAM)
Q8_0 Context 256K Speed ~101 tok/s
MoE — speed based on ~2.95 GB active experts plus per-token routing overhead (full 34.37 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~29 tok/s
Q8_0 Context 256K Speed ~17 tok/s
Q4_K_M Context 256K Speed ~130 tok/s
MoE — speed based on ~1.77 GB active experts plus per-token routing overhead (full 20.61 GB loaded in VRAM)
Q8_0 Context 256K Speed ~101 tok/s
MoE — speed based on ~2.95 GB active experts plus per-token routing overhead (full 34.37 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~65 tok/s
MoE — speed based on ~5.84 GB active experts plus per-token routing overhead (full 71.28 GB loaded in VRAM)
Q8_0 Context 256K Speed ~43 tok/s
MoE — speed based on ~9.91 GB active experts plus per-token routing overhead (full 120.95 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~27 tok/s
Q8_0 Context 128K Speed ~16 tok/s
Q4_K_M Context 16K Speed ~56 tok/s
Q8_0 Context 16K Speed ~34 tok/s
Q4_K_M Context 256K Speed ~32 tok/s
Q8_0 Context 256K Speed ~19 tok/s
Q4_K_M Context 256K Speed ~62 tok/s
MoE — speed based on ~6.24 GB active experts plus per-token routing overhead (full 111.7 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 195.2 GB
Q4_K_M Context 128K Speed ~13 tok/s
Q8_0 Context 128K Speed ~7 tok/s
Q4_K_M Context 128K Speed ~27 tok/s
Q8_0 Context 128K Speed ~16 tok/s
QwQ 32B 32.5B Excellent Reasoning Tool use MMLU 83
Q4_K_M Context 128K Speed ~27 tok/s
Q8_0 Context 128K Speed ~16 tok/s
Q4_K_M Context 118K Speed ~49 tok/s
MoE — speed based on ~8.51 GB active experts plus per-token routing overhead (full 175.8 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 306.8 GB
Q4_K_M Context 256K Speed ~112 tok/s
MoE — speed based on ~2.44 GB active experts plus per-token routing overhead (full 15.87 GB loaded in VRAM)
Q8_0 Context 256K Speed ~86 tok/s
MoE — speed based on ~3.85 GB active experts plus per-token routing overhead (full 25.02 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~84 tok/s
Q8_0 Context 256K Speed ~53 tok/s
Q4_K_M Context 192K Speed ~66 tok/s
MoE — speed based on ~5.63 GB active experts plus per-token routing overhead (full 128.84 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 226.7 GB
Q4_K_M Context 128K Speed ~134 tok/s
MoE — speed based on ~1.65 GB active experts plus per-token routing overhead (full 17.05 GB loaded in VRAM)
Q8_0 Context 128K Speed ~102 tok/s
MoE — speed based on ~2.87 GB active experts plus per-token routing overhead (full 29.66 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~37 tok/s
Q8_0 Context 128K Speed ~21 tok/s
Q4_K_M Context 256K Speed ~116 tok/s
MoE — speed based on ~2.25 GB active experts plus per-token routing overhead (full 23.73 GB loaded in VRAM)
Q8_0 Context 256K Speed ~98 tok/s
MoE — speed based on ~3.1 GB active experts plus per-token routing overhead (full 32.6 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~133 tok/s
MoE — speed based on ~1.7 GB active experts plus per-token routing overhead (full 17.28 GB loaded in VRAM)
Q8_0 Context 256K Speed ~100 tok/s
MoE — speed based on ~2.98 GB active experts plus per-token routing overhead (full 30.25 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~37 tok/s
Q8_0 Context 128K Speed ~21 tok/s
Q4_K_M Context 128K Speed ~27 tok/s
Q8_0 Context 128K Speed ~16 tok/s
Q4_K_M Context 128K Speed ~56 tok/s
Q8_0 Context 128K Speed ~34 tok/s
Q4_K_M Context 256K Speed ~115 tok/s
MoE — speed based on ~2.3 GB active experts plus per-token routing overhead (full 23.03 GB loaded in VRAM)
Q8_0 Context 256K Speed ~97 tok/s
MoE — speed based on ~3.13 GB active experts plus per-token routing overhead (full 31.28 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~44 tok/s
MoE — speed based on ~9.49 GB active experts plus per-token routing overhead (full 60.87 GB loaded in VRAM)
Q8_0 Context 128K Speed ~28 tok/s
MoE — speed based on ~16.64 GB active experts plus per-token routing overhead (full 106.67 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~150 tok/s
Q8_0 Context 256K Speed ~102 tok/s
Q4_K_M Context 128K Speed ~93 tok/s
Q8_0 Context 128K Speed ~58 tok/s
Q4_K_M Context 128K Speed ~27 tok/s
Q8_0 Context 128K Speed ~16 tok/s
Q4_K_M Context 128K Speed ~32 tok/s
Q8_0 Context 128K Speed ~19 tok/s
Q4_K_M Context 256K Speed ~69 tok/s
Q8_0 Context 256K Speed ~41 tok/s
Q4_K_M Context 128K Speed ~66 tok/s
Q8_0 Context 128K Speed ~40 tok/s
Q4_K_M Context 128K Speed ~127 tok/s
MoE — speed based on ~1.86 GB active experts plus per-token routing overhead (full 10.83 GB loaded in VRAM)
Q8_0 Context 128K Speed ~125 tok/s
MoE — speed based on ~1.93 GB active experts plus per-token routing overhead (full 11.28 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~99 tok/s
Q8_0 Context 128K Speed ~62 tok/s
Q4_K_M Context 128K Speed ~99 tok/s
Q8_0 Context 128K Speed ~62 tok/s
Q4_K_M Context 128K Speed ~67 tok/s
Q8_0 Context 128K Speed ~41 tok/s
Q4_K_M Context 128K Speed ~56 tok/s
Q8_0 Context 128K Speed ~34 tok/s
Q4_K_M Context 256K Speed ~88 tok/s
MoE — speed based on ~3.71 GB active experts plus per-token routing overhead (full 111.39 GB loaded in VRAM)
Q8_0 Context 256K Speed ~65 tok/s
MoE — speed based on ~5.84 GB active experts plus per-token routing overhead (full 175.33 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~76 tok/s
Q8_0 Context 128K Speed ~46 tok/s
Q4_K_M Context 128K Speed ~56 tok/s
Q8_0 Context 128K Speed ~34 tok/s
Q4_K_M Context 128K Speed ~37 tok/s
Q8_0 Context 128K Speed ~21 tok/s
Q4_K_M Context 128K Speed ~165 tok/s
Q8_0 Context 128K Speed ~111 tok/s
Q4_K_M Context 128K Speed ~160 tok/s
Q8_0 Context 128K Speed ~110 tok/s
Q4_K_M Context 128K Speed ~95 tok/s
Q8_0 Context 128K Speed ~59 tok/s
fp16 Context 128K Speed ~33 tok/s
Q4_K_M Context 128K Speed ~87 tok/s
Q8_0 Context 128K Speed ~62 tok/s
Q4_K_M Context 128K Speed ~66 tok/s
Q8_0 Context 128K Speed ~40 tok/s
Q4_K_M Context 128K Speed ~99 tok/s
Q8_0 Context 128K Speed ~62 tok/s
Q4_K_M Context 128K Speed ~191 tok/s
Q8_0 Context 128K Speed ~131 tok/s
Q4_K_M Context 32K Speed ~191 tok/s
Q8_0 Context 32K Speed ~131 tok/s
Q4_K_M Context 128K Speed ~87 tok/s
Q8_0 Context 128K Speed ~59 tok/s
Q4_K_M Context 32K Speed ~180 tok/s
Q8_0 Context 32K Speed ~124 tok/s
Q4_K_M Context 128K Speed ~186 tok/s
Q8_0 Context 128K Speed ~127 tok/s
fp16 Context 128K Speed ~75 tok/s
Q4_K_M Context 128K Speed ~160 tok/s
Q8_0 Context 128K Speed ~109 tok/s
Q4_K_M Context 32K Speed ~39 tok/s
Q8_0 Context 32K Speed ~23 tok/s
Q4_K_M Context 128K Speed ~126 tok/s
Q8_0 Context 128K Speed ~94 tok/s
Q4_K_M Context 128K Speed ~99 tok/s
Q8_0 Context 128K Speed ~62 tok/s
Q4_K_M Context 256K Speed ~246 tok/s
Q8_0 Context 256K Speed ~186 tok/s
Q4_K_M Context 16K Speed ~52 tok/s
Q8_0 Context 16K Speed ~31 tok/s
Q4_K_M Context 16K Speed ~110 tok/s
Q8_0 Context 16K Speed ~69 tok/s
Q4_K_M Context 128K Speed ~310 tok/s
Q8_0 Context 128K Speed ~241 tok/s
Q4_K_M Context 256K Speed ~364 tok/s
Q8_0 Context 256K Speed ~308 tok/s
Q4_K_M Context 1M Speed ~61 tok/s
MoE — speed based on ~6.3 GB active experts plus per-token routing overhead (full 65.29 GB loaded in VRAM)
Q8_0 Context 1M Speed ~38 tok/s
MoE — speed based on ~11.54 GB active experts plus per-token routing overhead (full 119.65 GB loaded in VRAM)
Q4_K_M Context 512K Speed ~85 tok/s
Q8_0 Context 512K Speed ~53 tok/s
Q4_K_M Context 512K Speed ~26 tok/s
Q8_0 Context 512K Speed ~15 tok/s
Q4_K_M Context 256K Speed ~141 tok/s
Q8_0 Context 256K Speed ~107 tok/s
Q4_K_S Context 256K Speed ~65 tok/s
MoE — speed based on ~5.77 GB active experts plus per-token routing overhead (full 103.84 GB loaded in VRAM)
Q8_0 Doesn't fit · needs 195.2 GB
Q4_K_M Context 256K Speed ~83 tok/s
Q8_0 Context 256K Speed ~52 tok/s
Q4_K_M Context 256K Speed ~133 tok/s
MoE — speed based on ~1.69 GB active experts plus per-token routing overhead (full 20.22 GB loaded in VRAM)
Q8_0 Context 256K Speed ~101 tok/s
MoE — speed based on ~2.94 GB active experts plus per-token routing overhead (full 35.21 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~132 tok/s
MoE — speed based on ~1.72 GB active experts plus per-token routing overhead (full 19.14 GB loaded in VRAM)
Q8_0 Context 256K Speed ~100 tok/s
MoE — speed based on ~2.97 GB active experts plus per-token routing overhead (full 33.15 GB loaded in VRAM)
Q8_0 Context 256K Speed ~47 tok/s
MoE — speed based on ~8.83 GB active experts plus per-token routing overhead (full 73.44 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~66 tok/s
MoE — speed based on ~5.65 GB active experts plus per-token routing overhead (full 70.64 GB loaded in VRAM)
Q8_0 Context 256K Speed ~43 tok/s
MoE — speed based on ~9.87 GB active experts plus per-token routing overhead (full 123.45 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~140 tok/s
MoE — speed based on ~1.48 GB active experts plus per-token routing overhead (full 23.61 GB loaded in VRAM)
Q8_0 Context 256K Speed ~111 tok/s
MoE — speed based on ~2.48 GB active experts plus per-token routing overhead (full 39.44 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~133 tok/s
MoE — speed based on ~1.68 GB active experts plus per-token routing overhead (full 19.71 GB loaded in VRAM)
Q8_0 Context 256K Speed ~101 tok/s
MoE — speed based on ~2.93 GB active experts plus per-token routing overhead (full 34.37 GB loaded in VRAM)
Q4_K_M Context 256K Speed ~82 tok/s
Q8_0 Context 256K Speed ~53 tok/s
Q4_K_M Context 256K Speed ~267 tok/s
Q8_0 Context 256K Speed ~197 tok/s
Arete 27B 27.36B Basic Vision Tool use Reasoning MMLU
Q4_K_M Context 256K Speed ~31 tok/s
Q8_0 Context 256K Speed ~19 tok/s
Q4_K_M Context 128K Speed ~32 tok/s
Q8_0 Context 128K Speed ~18 tok/s
Q4_K_M Context 128K Speed ~209 tok/s
Q8_0 Context 128K Speed ~144 tok/s
Q4_K_M Context 128K Speed ~144 tok/s
MoE — speed based on ~1.38 GB active experts plus per-token routing overhead (full 13.53 GB loaded in VRAM)
Q8_0 Context 128K Speed ~112 tok/s
MoE — speed based on ~2.43 GB active experts plus per-token routing overhead (full 23.74 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~138 tok/s
MoE — speed based on ~1.55 GB active experts plus per-token routing overhead (full 7.52 GB loaded in VRAM)
Q8_0 Context 128K Speed ~111 tok/s
MoE — speed based on ~2.48 GB active experts plus per-token routing overhead (full 12.04 GB loaded in VRAM)
Q4_K_M Context 128K Speed ~267 tok/s
Q8_0 Context 128K Speed ~197 tok/s
Q4_K_M Context 64K Speed ~78 tok/s
Q8_0 Context 64K Speed ~62 tok/s
Tight fit 2 Q4_K_M Context 8K Speed ~32 tok/s
Only 8K context — under the 16K needed to be practical
Q8_0 Context 8K Speed ~19 tok/s
Only 8K context — under the 16K needed to be practical
Q4_K_M Context 8K Speed ~83 tok/s
Only 8K context — under the 16K needed to be practical
Q8_0 Context 8K Speed ~52 tok/s
Only 8K context — under the 16K needed to be practical
Notify me when new models fit my GPU An email when a new model fits your M2 Ultra — no spam, unsubscribe anytime.
Doesn't fit 15 Q4_K_M Needs 381.2 GB — 189.2 GB over
Q8_0 Needs 664.4 GB — 472.4 GB over
Q4_K_M Needs 586.5 GB — 394.5 GB over
Q8_0 Needs 1013.0 GB — 821.0 GB over
Q4_K_M Needs 900.1 GB — 708.1 GB over
Q8_0 Needs 1670.1 GB — 1478.1 GB over
View all 15 →
Doesn't fit 15 ✕ Q4_K_M Needs 381.2 GB — 189.2 GB over
Q8_0 Needs 664.4 GB — 472.4 GB over
Q4_K_M Needs 586.5 GB — 394.5 GB over
Q8_0 Needs 1013.0 GB — 821.0 GB over
Q4_K_M Needs 900.1 GB — 708.1 GB over
Q8_0 Needs 1670.1 GB — 1478.1 GB over
Q4_K_M Needs 227.4 GB — 35.4 GB over
Q8_0 Needs 392.6 GB — 200.6 GB over
GLM-4.6 357B Excellent Reasoning Tool use MMLU 87
Q4_K_M Needs 200.8 GB — 8.8 GB over
Q8_0 Needs 353.3 GB — 161.3 GB over
Q4_K_M Needs 428.2 GB — 236.2 GB over
Q8_0 Needs 746.4 GB — 554.4 GB over
GLM 5.3 753.33B Excellent Tool use Reasoning MMLU 86.8
Q8_0 Needs 746.4 GB — 554.4 GB over
Q4_K_M Needs 226.3 GB — 34.3 GB over
Q8_0 Needs 396.8 GB — 204.8 GB over
GLM 5 753.86B Excellent Tool use MMLU 83
Q4_K_M Needs 424.7 GB — 232.7 GB over
Q8_0 Needs 746.4 GB — 554.4 GB over
Q4_K_M Needs 379.1 GB — 187.1 GB over
Q8_0 Needs 666.6 GB — 474.6 GB over
Q4_K_M Needs 591.3 GB — 399.3 GB over
Q8_0 Needs 1038.6 GB — 846.6 GB over
Q4_K_M Needs 465.3 GB — 273.3 GB over
Q8_0 Needs 811.9 GB — 619.9 GB over
Q4_K_M Needs 333.4 GB — 141.4 GB over
Q8_0 Needs 566.6 GB — 374.6 GB over
Q4_K_M Needs 293.0 GB — 101.0 GB over
Q8_0 Needs 496.2 GB — 304.2 GB over
Q4_K_M Needs 227.6 GB — 35.6 GB over
Q8_0 Needs 399.1 GB — 207.1 GB over