RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

What AI models run on the Mac M4 Max 128GB?

The Mac M4 Max 128GB has 128 GB of unified memory (546 GB/s). About 96.0 GB is usable for a model. 227 of 285 open models we track fit at 8K context.

227models fit (8K ctx)
227still fit at 32K ctx
241Blargest that fits
546 GB/smemory bandwidth

Models that fit, largest first

ModelParamsBest quant (8K)Needs32K contextMax tok/s (est.)
Intern S1241B (28B act.)Q2_K86.2 GBQ2_K52details
DeepSeek V2 Chat236B (21B act.)Q2_K83.5 GBQ2_K68details
MiniMax M2.5229B (11B act.)Q2_K80.3 GBQ2_K136details
MiniMax M2.7229B (11B act.)Q2_K82.6 GBQ2_K132details
MiniMax M2229B (11B act.)Q2_K80.3 GBQ2_K136details
MiniMax M2.1229B (11B act.)Q2_K80.3 GBQ2_K136details
MiniMax M3.1 Flash220B (4.4B act.)Q2_K78.7 GBQ2_K—details
Step 3.7 Flash201BQ2_K74.5 GBQ2_K7.2details
Step 3.5 Flash199BQ2_K69.0 GBQ2_K7.9details
Qwen3.8 Flash Next180B (62B act.)Q3_K_M83.0 GBQ3_K_M18details
Ling 3.0 flash Sante127B (5.6B act.)Q4_K_M75.9 GBQ4_K_M154details
Ling 3.0 flash127B (5.6B act.)Q5_K_M85.5 GBQ4_K_M137details
Ornith 1.5 122B125B (13B act.)Q5_K_M84.1 GBQ5_K_M60details
gpt oss 120b117B (5.8B act.)F1661.9 GBF16170details
gpt oss safeguard 120b117B (5.8B act.)Q8_060.1 GBQ8_0175details
GLM 4.5 Air110B (15B act.)Q5_K_M79.9 GBQ5_K_M48details
Llama 4 Scout 17B 16E Instruct109B (18B act.)Q6_K84.6 GBQ5_K_M37details
Llama 4 Scout 17B 16E109B (18B act.)Q6_K85.3 GBQ5_K_M37details
GLM 4.5V108B (14B act.)Q6_K84.4 GBQ5_K_M47details
GLM 4.6V108B (14B act.)Q5_K_M77.3 GBQ5_K_M51details
Hunyuan A13B Instruct80B (13B act.)Q8_081.2 GBQ8_040details
Kimi Dev 72B73BQ8_075.1 GBQ8_07.1details
Llama 3.3 70B Instruct71BQ8_073.0 GBQ8_07.3details
DeepSeek R1 Distill Llama 70B71BQ8_073.0 GBQ8_07.3details
Llama 3.1 70B71BQ8_073.0 GBQ8_07.3details
Kimi Linear 48B A3B Instruct49B (3.0B act.)Q8_049.6 GBQ8_0171details
Nous Hermes 2 Mixtral 8x7B DPO47B (13B act.)Q8_047.7 GBQ8_040details
falcon 40b42BF1679.4 GBF166.5details
flint40B (1.3B act.)F1675.1 GBF16218details
Maxwell V1 Flash37B (2.3B act.)F1669.7 GBF16120details
Hermes 4.3 36B36BF1670.0 GBF167.6details
Seed OSS 36B Instruct36BF1670.0 GBF167.6details
Qwen3.6 35B A3B36B (3.0B act.)F1667.2 GBF1692details
Ornith 1.5 35B A3B36B (3.0B act.)F1667.2 GBF1692details
Qwen3.5 35B A3B36B (3.0B act.)F1665.6 GBF1694details
Axion 1.0 35B35B (3.9B act.)F1665.6 GBF1671details
Qwen AgentWorld 35B A3B35B (3.0B act.)F1665.6 GBF1691details
Yi 34B 200K34BF1666.5 GBF167.9details
Yi 34B34BF1666.5 GBF167.9details
Yi 34B Chat34BF1666.5 GBF167.9details
Yi 1.5 34B Chat34BF1666.5 GBF167.9details
Nous Hermes 2 Yi 34B34BF1666.5 GBF167.9details
DeepSeek R1 Distill Qwen 32B33BF1663.7 GBF168.3details
Qwen3 32B33BF1663.7 GBF168.3details
GLM Z1 32B 041433BF1661.8 GBF168.4details
GLM 4 32B 041433BF1661.8 GBF168.4details
granite 4.0 h small32B (8.8B act.)F1660.7 GBF1631details
gemma 4 31B it31BF1660.1 GBF168.9details
GLM 4.7 Flash31B (4.6B act.)F1656.9 GBF1662details
Qwen3 Coder 30B A3B Instruct31B (3.0B act.)F1658.3 GBF1691details
North Mini Code 1.030B (3.3B act.)F1658.2 GBF1683details
Hy Memory 30B A3B30B (3.0B act.)F1657.4 GBF1691details
Muse Glimmer 30B30BF1652.9 GBF169.8details
granite 4.2 30b29BF1657.1 GBF169.3details
Qwen3.8 27B28BF1652.3 GBF1610.0details
Qwen3.6 27B28BF1651.5 GBF1610details
Qwen3.5 27B28BF1651.5 GBF1610details
gemma 3 27b it27BF1652.2 GBF1610details
Arete 27B27BF1651.5 GBF1610details
MiniMax H3 Image PE27BF1652.3 GBF1610.0details
Trinity Mini26B (2.0B act.)F1649.7 GBF16139details
GLiDE 27B26BF1650.0 GBF1610details
diffusiongemma 26B A4B it26B (4.0B act.)F1648.5 GBF1670details
gemma 4 26B A4B it26B (4.0B act.)F1648.4 GBF1670details
LFM2.5 24B A2B24B (2.0B act.)Q8_024.5 GBQ8_0257details
ERNIE 4.5 21B A3B PT22BF1641.7 GBF1613details
ERNIE 4.5 21B A3B Thinking22BF1641.7 GBF1613details
gpt oss safeguard 20b22B (4.8B act.)F1613.8 GBF16178details
gpt oss 20b21B (4.2B act.)F1613.8 GBF16197details
Groot 19B A3B 26102319B (3.0B act.)F1636.4 GBF1693details
Kimi VL A3B Thinking 250616B (3.0B act.)F1660.4 GBF1647details
Kimi VL A3B Instruct16B (3.0B act.)F1631.5 GBF1691details
Kimi VL A3B Thinking16B (3.0B act.)F1631.5 GBF1691details
deepseek moe 16b base16B (2.8B act.)F1632.8 GBF1697details
LLaDA2.1 mini16B (1.4B act.)F1631.3 GBF16191details
Moonlight 16B A3B Instruct16B (3.0B act.)F1630.7 GBF1691details
DeepSeek Coder V2 Lite Instruct16B (2.7B act.)F1630.1 GBF16103details
DeepSeek V2 Lite16B (2.7B act.)F1630.1 GBF16103details
DeepSeek V2 Lite Chat16B (2.7B act.)F1630.1 GBF16103details
Qwen2.5 Coder 14B Instruct15BF1629.7 GBF1618details
DeepSeek R1 Distill Qwen 14B15BF1629.7 GBF1618details
Qwen3 14B15BF1629.4 GBF1618details
phi 415BF1629.5 GBF1619details
gemma 3 12b it12BF1623.6 GBF1623details
Llama Guard 4 12B12BF1624.6 GBF1623details
Llama 3.2 11B Vision Instruct11BF1620.1 GBF1628details
Llama 3.2 11B Vision11BF1621.7 GBF1626details
GLM 4.6V Flash10BF1618.5 GBF1629details
GLM 4.1V 9B Thinking10BF1618.5 GBF1629details
Step3 VL 10B10BF1632.3 GBF1617details
SenseNova Loop 9B Preview9.7BF1617.8 GBF1630details
Qwen3.5 9B9.7BF1617.8 GBF1630details
Ornith 1.5 9B9.7BF1617.8 GBF1630details
Qwen3.5 9B Base9.7BF1617.8 GBF1630details
MiMo V2.6 Distill Qwen 9B9.4BF1617.8 GBF1630details
glm 4 9b9.4BF1623.2 GBF1629details
internlm3 8b instruct8.8BF1617.4 GBF1631details
granite 4.2 8b8.8BF1618.2 GBF1631details
granite 4.1 8b8.8BF1618.2 GBF1631details
Qwen3 VL 8B Instruct8.8BF1617.1 GBF1633details
Molmo2 8B8.7BF1617.1 GBF1633details
LFM2.5 8B A1B8.5B (1.0B act.)F1632.3 GBF16136details
Qwen2.5 VL 7B Instruct8.3BF1615.3 GBF1636details
UI TARS 1.5 7B8.3BF1615.3 GBF1636details
olmOCR 2 7B 10258.3BF1615.3 GBF1636details
olmOCR 7B 0225 preview8.3BF1616.6 GBF1633details
UI TARS 7B DPO8.3BF1615.3 GBF1636details
Qwen3 8B8.2BF1617.1 GBF1633details
DeepSeek R1 0528 Qwen3 8B8.2BF1617.1 GBF1633details
granite 3.3 8b instruct8.2BF1617.0 GBF1633details
granite 3.1 8b instruct8.2BF1617.0 GBF1633details
granite 3.2 8b instruct8.2BF1617.0 GBF1633details
Apertus 8B Instruct 25098.1BF1616.7 GBF1634details
Llama 3.1 8B Instruct8.0BF1616.6 GBF1634details
Llama 3.1 8B8.0BF1616.6 GBF1634details
Hermes 3 Llama 3.1 8B8.0BF1616.6 GBF1634details
DeepSeek R1 Distill Llama 8B8.0BF1631.6 GBF1617details
Meta Llama 3 8B Instruct8.0BF1616.6 GBF1634details
Meta Llama Guard 2 8B8.0BF1616.6 GBF1634details
Meta Llama 3.1 8B Instruct8.0BF1616.6 GBF1634details
Ling 3.0 tiny7.9B (1.4B act.)F1615.6 GBF16197details
MiMo 7B RL7.8BF1616.4 GBF1635details
internlm2 5 7b chat7.7BF1616.0 GBF1635details
internlm2 chat 7b7.7BF1616.0 GBF1635details
Qwen2.5 7B Instruct7.6BF1615.3 GBF1636details
Qwen2.5 Coder 7B Instruct7.6BF1615.3 GBF1636details
DeepSeek R1 Distill Qwen 7B7.6BF1615.3 GBF1636details
Falcon H1R 7B7.6BF1615.1 GBF1636details
Olmo 3 7B Instruct7.3BF1616.7 GBF1637details
Olmo 3 1025 7B7.3BF1616.7 GBF1637details
Olmo 3 7B Think7.3BF1616.7 GBF1637details
Mistral 7B v0.17.2BF1614.9 GBF1638details
Mistral 7B Instruct v0.27.2BF1614.9 GBF1638details
Mistral 7B Instruct v0.17.2BF1614.9 GBF1638details
falcon 7b7.2BF1618.4 GBF1638details
falcon 7b instruct7.2BF1618.4 GBF1638details
granite 4.0 h tiny6.9B (1.5B act.)F1613.6 GBF16186details
OLMoE 1B 7B 09246.9B (1.3B act.)F1614.4 GBF16213details
deepseek coder 7b instruct v1.56.9BF1617.2 GBF1640details
deepseek coder 6.7b instruct6.7BF1617.0 GBF1641details
deepseek coder 6.7b base6.7BF1617.0 GBF1641details
Llama 2 7b hf6.7BF1617.0 GBF1641details
Yi 6B 200K6.1BF1612.3 GBF1645details
Yi 6B6.1BF1612.3 GBF1645details
Molmo2 4B4.9BF1610.0 GBF1662details
WeVisOne VL 4B4.8BF1610.0 GBF1662details
Mage VL Cowork4.7BF1610.0 GBF1662details
Qianfan OCR4.7BF1610.1 GBF1662details
Qwen3.5 4B4.7BF169.0 GBF1665details
S5 nano4.7BF169.2 GBF1663details
LightOnOCR 3 4B4.5BF169.0 GBF1665details
SingVL S1 4B4.5BF169.0 GBF1665details
Intern Decision 4B4.5BF169.0 GBF1665details
Qwen3 VL 4B Instruct4.4BF169.3 GBF1668details
gemma 3 4b it4.3BF168.4 GBF1670details
medgemma 1.5 4b it4.3BF168.4 GBF1670details
medgemma 4b it4.3BF168.4 GBF1670details
gemma 3 4b pt4.3BF169.2 GBF1663details
Phi 3.5 vision instruct4.1BF1611.2 GBF1666details
Qwen3 4B4.0BF169.3 GBF1668details
Qwen3 4B Instruct 25074.0BF169.3 GBF1668details
granite vision 4.1 4b4.0BF167.6 GBF1680details
Phi 4 mini instruct3.8BF168.9 GBF1671details
Phi 3 mini 4k instruct3.8BF1610.6 GBF1671details
Phi 3.5 mini instruct3.8BF1610.6 GBF1671details
Phi 3 mini 128k instruct3.8BF1610.6 GBF1671details
Qwen2.5 VL 3B Instruct3.8BF166.7 GBF1688details
granite 4.2 3b3.7BF168.0 GBF1675details
granite 4.1 3b llmon preview3.7BF168.0 GBF1675details
Raon 1 nano3.5BF167.0 GBF1687details
granite 4.1 3b3.4BF167.6 GBF1680details
granite 4.0 micro3.4BF167.6 GBF1680details
DeepSeek OCR 23.4B (1.2B act.)F166.6 GBF16264details
deepseek vl2 tiny3.4B (1.2B act.)F167.4 GBF16232details
Unlimited OCR 23.3B (1.1B act.)F166.6 GBF16270details
Unlimited OCR3.3B (1.1B act.)F166.6 GBF16272details
DeepSeek OCR3.3B (1.1B act.)F166.6 GBF16272details
granite 3.1 3b a800m instruct3.3B (883M act.)F167.1 GBF16309details
Llama 3.2 3B Instruct3.2BF1613.5 GBF1642details
Llama 3.2 3B3.2BF167.5 GBF1685details
Hermes 3 Llama 3.2 3B3.2BF167.5 GBF1685details
d1 3B3.1BF1610.8 GBF1651details
LFM2.5 VL 3B3.1BF1610.8 GBF1651details
Qwen2.5 3B Instruct3.1BF166.7 GBF1688details
SmolLM3 3B3.1BF166.9 GBF1689details
SmolLM3 3B Base3.1BF166.9 GBF1689details
phi 22.8BF168.2 GBF1698details
LFM2.5 2.6B2.7BF1610.8 GBF1651details
granite 3.3 2b instruct2.5BF165.8 GBF16108details
MiniCPM5 2B2.5BF165.7 GBF16108details
WeVisOne VL 2B2.4BF165.4 GBF16134details
Qwen3.5 2B2.3BF164.5 GBF16145details
SmolVLM Instruct2.2BF165.4 GBF16151details
Qwen3 1.7B2.0BF164.8 GBF16158details
Qwen3.8 27B DSpine b162.0BF164.8 GBF16134details
d1 2B1.9BF167.2 GBF1679details
LFM3 VL 2B1.9BF167.2 GBF1679details
DeepSeek R1 Distill Qwen 1.5B1.8BF164.2 GBF16153details
LFM3 1.7B1.7BF167.2 GBF1679details
LFM3 1.7B Base1.7BF164.0 GBF16158details
SmolLM2 1.7B Instruct1.7BF165.2 GBF16159details
SmolLM2 1.7B1.7BF165.2 GBF16160details
SmolLM 1.7B1.7BF165.2 GBF16159details
granite 4.0 1b1.6BF164.3 GBF16167details
Qwen2.5 1.5B Instruct1.5BF163.8 GBF16177details
OLMo 2 0425 1B1.5BF164.4 GBF16184details
deepseek coder 1.3b instruct1.3BF164.5 GBF16203details
Llama 3.2 1B1.2BF163.2 GBF16221details
Llama 3.2 1B Instruct1.2BF165.5 GBF16110details
LFM2.5 1.2B Instruct1.2BF165.0 GBF16117details
HunyuanOCR1.1BF163.5 GBF16244details
PaddleOCR VL 1.6959MF162.5 GBF16285details
Qwen3.5 0.8B873MF162.4 GBF16360details
LightOnOCR 3 0.8B853MF162.4 GBF16360details
FastVLM 0.5B759MF162.2 GBF16360details
Qwen3 0.6B752MF162.7 GBF16456details
GOT OCR2 0716MF162.8 GBF16381details
Nemotron 3.5 Parse631MF161.9 GBF16418details
GOT OCR 2.0 hf561MF162.5 GBF16487details
SmolVLM2 500M Video Instruct507MF161.6 GBF16666details
SmolVLM 500M Instruct507MF161.6 GBF16666details
Qwen2.5 0.5B Instruct494MF161.7 GBF16553details
LFM2.5 VL 450M449MF161.9 GBF16384details
SmolLM2 360M Instruct362MF161.5 GBF16753details
SmolLM2 360M362MF161.5 GBF16753details
LFM2.5 350M354MF161.9 GBF16384details
LFM2 350M354MF161.4 GBF16770details

Just too big at 4-bit

These need up to 35% more memory than the Mac M4 Max 128GB has at Q4_K_M and 8K context. A smaller quant or shorter context may work.

ModelQ4_K_M needsShort by
Qwen3.8 Flash Next103.5 GB7.5 GB
Step 3.5 Flash115.9 GB19.9 GB
Step 3.7 Flash118.8 GB22.8 GB
MiniMax M3.1 Flash127.2 GB31.2 GB

Questions

What is the biggest model that runs on the Mac M4 Max 128GB?

Of the models we track, the largest that fits at 8K context is Intern S1 (241B) at Q2_K, needing about 86.2 GB.

How much of the Mac M4 Max 128GB's memory can a model use?

About 96 GB of 128 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.