RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

What AI models run on the Mac M2 Ultra 192GB?

The Mac M2 Ultra 192GB has 192 GB of unified memory (800 GB/s). About 144 GB is usable for a model. 245 of 285 open models we track fit at 8K context.

245models fit (8K ctx)
244still fit at 32K ctx
402Blargest that fits
800 GB/smemory bandwidth

Apple Silicon uses unified memory; see the Mac guide for default macOS limits across machines.

Models that fit, largest first

ModelParamsBest quant (8K)Needs32K contextMax tok/s (est.)
Llama 4 Maverick 17B 128E Instruct402B (18B act.)Q2_K138 GBQ2_K122details
GLM 4.5358B (36B act.)Q2_K126 GBQ2_K61details
GLM 4.7358B (36B act.)Q2_K126 GBQ2_K61details
GLM 4.6357B (34B act.)Q2_K125 GBQ2_K64details
GLM 5.3 Flash321B (18B act.)Q2_K113 GBQ2_K116details
Nex N2.6321B (18B act.)Q2_K118 GBQ2_K112details
step3321BQ2_K143 GBno6.6details
MiMo V2.6 Flash RL311B (6.2B act.)Q2_K110 GBQ2_K—details
MiMo V2.5311B (17B act.)Q2_K110 GBQ2_K122details
MiMo V2.6 Flash MOPD311B (6.2B act.)Q2_K110 GBQ2_K—details
MiMo V2 Flash310B (16B act.)Q2_K106 GBQ2_K134details
DeepSeek V4 Flash Vision Exp305B (15B act.)Q2_K108 GBQ2_K140details
DeepSeek V4 Flash 0731304B (27B act.)Q2_K108 GBQ2_K78details
DeepSeek V4 Flash DSpark304B (27B act.)Q2_K108 GBQ2_K78details
Hy3299B (21B act.)Q2_K102 GBQ2_K107details
Hy3 preview299B (21B act.)Q2_K107 GBQ2_K102details
DeepSeek V4 Flash291B (14B act.)Q2_K103 GBQ2_K151details
ERNIE 5.5 Flash Base286B (5.7B act.)Q2_K101 GBQ2_K—details
Intern S1241B (28B act.)Q3_K_M112 GBQ3_K_M59details
DeepSeek V2 Chat236B (21B act.)Q3_K_M108 GBQ3_K_M77details
MiniMax M2.5229B (11B act.)Q3_K_M104 GBQ3_K_M152details
MiniMax M2.7229B (11B act.)Q3_K_M97.0 GBQ3_K_M164details
MiniMax M2229B (11B act.)Q3_K_M104 GBQ3_K_M152details
MiniMax M2.1229B (11B act.)Q3_K_M104 GBQ3_K_M152details
MiniMax M3.1 Flash220B (4.4B act.)Q4_K_M127 GBQ3_K_M—details
Step 3.7 Flash201BQ4_K_M119 GBQ4_K_M6.5details
Step 3.5 Flash199BQ4_K_M116 GBQ4_K_M6.7details
Qwen3.8 Flash Next180B (62B act.)Q5_K_M121 GBQ5_K_M18details
Ling 3.0 flash Sante127B (5.6B act.)Q8_0127 GBQ8_0134details
Ling 3.0 flash127B (5.6B act.)Q8_0127 GBQ8_0134details
Ornith 1.5 122B125B (13B act.)Q8_0125 GBQ8_059details
gpt oss 120b117B (5.8B act.)F1661.9 GBF16249details
gpt oss safeguard 120b117B (5.8B act.)Q8_060.1 GBQ8_0256details
GLM 4.5 Air110B (15B act.)Q8_0112 GBQ8_050details
Llama 4 Scout 17B 16E Instruct109B (18B act.)Q8_0109 GBQ8_042details
Llama 4 Scout 17B 16E109B (18B act.)Q8_0110 GBQ8_042details
GLM 4.5V108B (14B act.)Q8_0108 GBQ8_053details
GLM 4.6V108B (14B act.)Q8_0108 GBQ8_053details
Hunyuan A13B Instruct80B (13B act.)Q8_081.2 GBQ8_058details
Kimi Dev 72B73BQ8_075.1 GBQ8_010details
Llama 3.3 70B Instruct71BQ8_073.0 GBQ8_011details
DeepSeek R1 Distill Llama 70B71BQ8_073.0 GBQ8_011details
Llama 3.1 70B71BQ8_073.0 GBQ8_011details
Kimi Linear 48B A3B Instruct49B (3.0B act.)F1692.5 GBF16133details
Nous Hermes 2 Mixtral 8x7B DPO47B (13B act.)F1688.5 GBF1631details
falcon 40b42BF1679.4 GBF169.6details
flint40B (1.3B act.)F1675.1 GBF16320details
Maxwell V1 Flash37B (2.3B act.)F1669.7 GBF16176details
Hermes 4.3 36B36BF1670.0 GBF1611details
Seed OSS 36B Instruct36BF1670.0 GBF1611details
Qwen3.6 35B A3B36B (3.0B act.)F1667.2 GBF16135details
Ornith 1.5 35B A3B36B (3.0B act.)F1667.2 GBF16135details
Qwen3.5 35B A3B36B (3.0B act.)F1665.6 GBF16138details
Axion 1.0 35B35B (3.9B act.)F1665.6 GBF16104details
Qwen AgentWorld 35B A3B35B (3.0B act.)F1665.6 GBF16133details
Yi 34B 200K34BF1666.5 GBF1612details
Yi 34B34BF1666.5 GBF1612details
Yi 34B Chat34BF1666.5 GBF1612details
Yi 1.5 34B Chat34BF1666.5 GBF1612details
Nous Hermes 2 Yi 34B34BF1666.5 GBF1612details
DeepSeek R1 Distill Qwen 32B33BF1663.7 GBF1612details
Qwen3 32B33BF1663.7 GBF1612details
GLM Z1 32B 041433BF1661.8 GBF1612details
GLM 4 32B 041433BF1661.8 GBF1612details
granite 4.0 h small32B (8.8B act.)F1660.7 GBF1645details
gemma 4 31B it31BF1660.1 GBF1613details
GLM 4.7 Flash31B (4.6B act.)F1656.9 GBF1690details
Qwen3 Coder 30B A3B Instruct31B (3.0B act.)F1658.3 GBF16133details
North Mini Code 1.030B (3.3B act.)F1658.2 GBF16121details
Hy Memory 30B A3B30B (3.0B act.)F1657.4 GBF16133details
Muse Glimmer 30B30BF1652.9 GBF1614details
granite 4.2 30b29BF1657.1 GBF1614details
Qwen3.8 27B28BF1652.3 GBF1615details
Qwen3.6 27B28BF1651.5 GBF1615details
Qwen3.5 27B28BF1651.5 GBF1615details
gemma 3 27b it27BF1652.2 GBF1615details
Arete 27B27BF1651.5 GBF1615details
MiniMax H3 Image PE27BF1652.3 GBF1615details
Trinity Mini26B (2.0B act.)F1649.7 GBF16204details
GLiDE 27B26BF1650.0 GBF1615details
diffusiongemma 26B A4B it26B (4.0B act.)F1648.5 GBF16102details
gemma 4 26B A4B it26B (4.0B act.)F1648.4 GBF16102details
LFM2.5 24B A2B24B (2.0B act.)F1690.1 GBF16100details
ERNIE 4.5 21B A3B PT22BF1641.7 GBF1618details
ERNIE 4.5 21B A3B Thinking22BF1641.7 GBF1618details
gpt oss safeguard 20b22B (4.8B act.)F1613.8 GBF16260details
gpt oss 20b21B (4.2B act.)F1613.8 GBF16289details
Groot 19B A3B 26102319B (3.0B act.)F1636.4 GBF16136details
Kimi VL A3B Thinking 250616B (3.0B act.)F1660.4 GBF1669details
Kimi VL A3B Instruct16B (3.0B act.)F1631.5 GBF16133details
Kimi VL A3B Thinking16B (3.0B act.)F1631.5 GBF16133details
deepseek moe 16b base16B (2.8B act.)F1632.8 GBF16141details
LLaDA2.1 mini16B (1.4B act.)F1631.3 GBF16279details
Moonlight 16B A3B Instruct16B (3.0B act.)F1630.7 GBF16133details
DeepSeek Coder V2 Lite Instruct16B (2.7B act.)F1630.1 GBF16150details
DeepSeek V2 Lite16B (2.7B act.)F1630.1 GBF16150details
DeepSeek V2 Lite Chat16B (2.7B act.)F1630.1 GBF16150details
Qwen2.5 Coder 14B Instruct15BF1629.7 GBF1627details
DeepSeek R1 Distill Qwen 14B15BF1629.7 GBF1627details
Qwen3 14B15BF1629.4 GBF1627details
phi 415BF1629.5 GBF1627details
gemma 3 12b it12BF1623.6 GBF1634details
Llama Guard 4 12B12BF1624.6 GBF1633details
Llama 3.2 11B Vision Instruct11BF1620.1 GBF1641details
Llama 3.2 11B Vision11BF1621.7 GBF1638details
GLM 4.6V Flash10BF1618.5 GBF1643details
GLM 4.1V 9B Thinking10BF1618.5 GBF1643details
Step3 VL 10B10BF1632.3 GBF1624details
SenseNova Loop 9B Preview9.7BF1617.8 GBF1645details
Qwen3.5 9B9.7BF1617.8 GBF1645details
Ornith 1.5 9B9.7BF1617.8 GBF1645details
Qwen3.5 9B Base9.7BF1617.8 GBF1645details
MiMo V2.6 Distill Qwen 9B9.4BF1617.8 GBF1645details
glm 4 9b9.4BF1623.2 GBF1643details
internlm3 8b instruct8.8BF1617.4 GBF1645details
granite 4.2 8b8.8BF1618.2 GBF1645details
granite 4.1 8b8.8BF1618.2 GBF1645details
Qwen3 VL 8B Instruct8.8BF1617.1 GBF1649details
Molmo2 8B8.7BF1617.1 GBF1649details
LFM2.5 8B A1B8.5B (1.0B act.)F1632.3 GBF16200details
Qwen2.5 VL 7B Instruct8.3BF1615.3 GBF1653details
UI TARS 1.5 7B8.3BF1615.3 GBF1653details
olmOCR 2 7B 10258.3BF1615.3 GBF1653details
olmOCR 7B 0225 preview8.3BF1616.6 GBF1648details
UI TARS 7B DPO8.3BF1615.3 GBF1653details
Qwen3 8B8.2BF1617.1 GBF1649details
DeepSeek R1 0528 Qwen3 8B8.2BF1617.1 GBF1649details
granite 3.3 8b instruct8.2BF1617.0 GBF1649details
granite 3.1 8b instruct8.2BF1617.0 GBF1649details
granite 3.2 8b instruct8.2BF1617.0 GBF1649details
Apertus 8B Instruct 25098.1BF1616.7 GBF1650details
Llama 3.1 8B Instruct8.0BF1616.6 GBF1650details
Llama 3.1 8B8.0BF1616.6 GBF1650details
Hermes 3 Llama 3.1 8B8.0BF1616.6 GBF1650details
DeepSeek R1 Distill Llama 8B8.0BF1631.6 GBF1625details
Meta Llama 3 8B Instruct8.0BF1616.6 GBF1650details
Meta Llama Guard 2 8B8.0BF1616.6 GBF1650details
Meta Llama 3.1 8B Instruct8.0BF1616.6 GBF1650details
Ling 3.0 tiny7.9B (1.4B act.)F1615.6 GBF16289details
MiMo 7B RL7.8BF1616.4 GBF1651details
internlm2 5 7b chat7.7BF1616.0 GBF1652details
internlm2 chat 7b7.7BF1616.0 GBF1652details
Qwen2.5 7B Instruct7.6BF1615.3 GBF1653details
Qwen2.5 Coder 7B Instruct7.6BF1615.3 GBF1653details
DeepSeek R1 Distill Qwen 7B7.6BF1615.3 GBF1653details
Falcon H1R 7B7.6BF1615.1 GBF1653details
Olmo 3 7B Instruct7.3BF1616.7 GBF1655details
Olmo 3 1025 7B7.3BF1616.7 GBF1655details
Olmo 3 7B Think7.3BF1616.7 GBF1655details
Mistral 7B v0.17.2BF1614.9 GBF1655details
Mistral 7B Instruct v0.27.2BF1614.9 GBF1655details
Mistral 7B Instruct v0.17.2BF1614.9 GBF1655details
falcon 7b7.2BF1618.4 GBF1655details
falcon 7b instruct7.2BF1618.4 GBF1655details
granite 4.0 h tiny6.9B (1.5B act.)F1613.6 GBF16273details
OLMoE 1B 7B 09246.9B (1.3B act.)F1614.4 GBF16312details
deepseek coder 7b instruct v1.56.9BF1617.2 GBF1658details
deepseek coder 6.7b instruct6.7BF1617.0 GBF1659details
deepseek coder 6.7b base6.7BF1617.0 GBF1659details
Llama 2 7b hf6.7BF1617.0 GBF1659details
Yi 6B 200K6.1BF1612.3 GBF1666details
Yi 6B6.1BF1612.3 GBF1666details
Molmo2 4B4.9BF1610.0 GBF1691details
WeVisOne VL 4B4.8BF1610.0 GBF1691details
Mage VL Cowork4.7BF1610.0 GBF1691details
Qianfan OCR4.7BF1610.1 GBF1690details
Qwen3.5 4B4.7BF169.0 GBF1695details
S5 nano4.7BF169.2 GBF1692details
LightOnOCR 3 4B4.5BF169.0 GBF1695details
SingVL S1 4B4.5BF169.0 GBF1695details
Intern Decision 4B4.5BF169.0 GBF1695details
Qwen3 VL 4B Instruct4.4BF169.3 GBF1699details
gemma 3 4b it4.3BF168.4 GBF16103details
medgemma 1.5 4b it4.3BF168.4 GBF16103details
medgemma 4b it4.3BF168.4 GBF16103details
gemma 3 4b pt4.3BF169.2 GBF1693details
Phi 3.5 vision instruct4.1BF1611.2 GBF1696details
Qwen3 4B4.0BF169.3 GBF1699details
Qwen3 4B Instruct 25074.0BF169.3 GBF1699details
granite vision 4.1 4b4.0BF167.6 GBF16117details
Phi 4 mini instruct3.8BF168.9 GBF16104details
Phi 3 mini 4k instruct3.8BF1610.6 GBF16105details
Phi 3.5 mini instruct3.8BF1610.6 GBF16105details
Phi 3 mini 128k instruct3.8BF1610.6 GBF16105details
Qwen2.5 VL 3B Instruct3.8BF166.7 GBF16129details
granite 4.2 3b3.7BF168.0 GBF16109details
granite 4.1 3b llmon preview3.7BF168.0 GBF16109details
Raon 1 nano3.5BF167.0 GBF16127details
granite 4.1 3b3.4BF167.6 GBF16117details
granite 4.0 micro3.4BF167.6 GBF16117details
DeepSeek OCR 23.4B (1.2B act.)F166.6 GBF16386details
deepseek vl2 tiny3.4B (1.2B act.)F167.4 GBF16340details
Unlimited OCR 23.3B (1.1B act.)F166.6 GBF16396details
Unlimited OCR3.3B (1.1B act.)F166.6 GBF16398details
DeepSeek OCR3.3B (1.1B act.)F166.6 GBF16398details
granite 3.1 3b a800m instruct3.3B (883M act.)F167.1 GBF16453details
Llama 3.2 3B Instruct3.2BF1613.5 GBF1662details
Llama 3.2 3B3.2BF167.5 GBF16125details
Hermes 3 Llama 3.2 3B3.2BF167.5 GBF16124details
d1 3B3.1BF1610.8 GBF1674details
LFM2.5 VL 3B3.1BF1610.8 GBF1674details
Qwen2.5 3B Instruct3.1BF166.7 GBF16130details
SmolLM3 3B3.1BF166.9 GBF16130details
SmolLM3 3B Base3.1BF166.9 GBF16130details
phi 22.8BF168.2 GBF16144details
LFM2.5 2.6B2.7BF1610.8 GBF1674details
granite 3.3 2b instruct2.5BF165.8 GBF16158details
MiniCPM5 2B2.5BF165.7 GBF16159details
WeVisOne VL 2B2.4BF165.4 GBF16196details
Qwen3.5 2B2.3BF164.5 GBF16212details
SmolVLM Instruct2.2BF165.4 GBF16221details
Qwen3 1.7B2.0BF164.8 GBF16232details
Qwen3.8 27B DSpine b162.0BF164.8 GBF16197details
d1 2B1.9BF167.2 GBF16115details
LFM3 VL 2B1.9BF167.2 GBF16115details
DeepSeek R1 Distill Qwen 1.5B1.8BF164.2 GBF16225details
LFM3 1.7B1.7BF167.2 GBF16115details
LFM3 1.7B Base1.7BF164.0 GBF16231details
SmolLM2 1.7B Instruct1.7BF165.2 GBF16234details
SmolLM2 1.7B1.7BF165.2 GBF16234details
SmolLM 1.7B1.7BF165.2 GBF16234details
granite 4.0 1b1.6BF164.3 GBF16245details
Qwen2.5 1.5B Instruct1.5BF163.8 GBF16259details
OLMo 2 0425 1B1.5BF164.4 GBF16269details
deepseek coder 1.3b instruct1.3BF164.5 GBF16297details
Llama 3.2 1B1.2BF163.2 GBF16324details
Llama 3.2 1B Instruct1.2BF165.5 GBF16161details
LFM2.5 1.2B Instruct1.2BF165.0 GBF16171details
HunyuanOCR1.1BF163.5 GBF16357details
PaddleOCR VL 1.6959MF162.5 GBF16417details
Qwen3.5 0.8B873MF162.4 GBF16527details
LightOnOCR 3 0.8B853MF162.4 GBF16527details
FastVLM 0.5B759MF162.2 GBF16527details
Qwen3 0.6B752MF162.7 GBF16668details
GOT OCR2 0716MF162.8 GBF16559details
Nemotron 3.5 Parse631MF161.9 GBF16612details
GOT OCR 2.0 hf561MF162.5 GBF16714details
SmolVLM2 500M Video Instruct507MF161.6 GBF16975details
SmolVLM 500M Instruct507MF161.6 GBF16975details
Qwen2.5 0.5B Instruct494MF161.7 GBF16810details
LFM2.5 VL 450M449MF161.9 GBF16562details
SmolLM2 360M Instruct362MF161.5 GBF161103details
SmolLM2 360M362MF161.5 GBF161103details
LFM2.5 350M354MF161.9 GBF16562details
LFM2 350M354MF161.4 GBF161128details

Just too big at 4-bit

These need up to 35% more memory than the Mac M2 Ultra 192GB has at Q4_K_M and 8K context. A smaller quant or shorter context may work.

ModelQ4_K_M needsShort by
ERNIE 5.5 Flash Base164.2 GB20.2 GB
DeepSeek V4 Flash167.0 GB23.0 GB
Hy3172.8 GB28.8 GB
Hy3 preview173.2 GB29.2 GB
DeepSeek V4 Flash 0731174.6 GB30.6 GB
DeepSeek V4 Flash DSpark174.6 GB30.6 GB
DeepSeek V4 Flash Vision Exp174.8 GB30.8 GB
MiMo V2 Flash175.8 GB31.8 GB
MiMo V2.6 Flash RL178.7 GB34.7 GB
MiMo V2.6 Flash MOPD178.7 GB34.7 GB
MiMo V2.5179.6 GB35.6 GB
GLM 5.3 Flash183.7 GB39.7 GB
Nex N2.6187.8 GB43.8 GB

Questions

What is the biggest model that runs on the Mac M2 Ultra 192GB?

Of the models we track, the largest that fits at 8K context is Llama 4 Maverick 17B 128E Instruct (402B) at Q2_K, needing about 138 GB.

How much of the Mac M2 Ultra 192GB's memory can a model use?

About 144 GB of 192 GB. macOS lets the GPU use about 75% of unified memory by default; this can be raised with the iogpu.wired_limit_mb setting.

Spec source: Apple specifications. Model data: Hugging Face Hub API. Speed is a bandwidth ceiling (bandwidth ÷ bytes read per token), not a benchmark. Methodology · Report an error.