Which AI models can the Apple M4 Max (64GB) run?
Every model we track, checked against the Apple M4 Max (64GB) at 8K context. The largest that runs well is Qwen2.5 72B.
Unified memory64 GB48 GB usable by the GPU
Bandwidth546GB/s
Runs well53/69models at Q4_K_M or better
Fastest model~223tok/s · Qwen3 Coder Next 80B-A3B
Every model on the Apple M4 Max (64GB)
Bars show memory used out of 48 GB. Pick a row for the full breakdown.
53 Runs well
NameMemory usedBest quantSpeed
Qwen2.5 72B73BQ4_K_M · 47.4 GB~7 tok/s
Llama 3.1 70B71BQ4_K_M · 46.1 GB~8 tok/s- Llama 3.3 70B71BQ4_K_M · 46.1 GB~8 tok/s
DeepSeek R1 Distill Llama 70B71BQ4_K_M · 46.1 GB~8 tok/s
Mixtral 8x7B47BQ6_K · 40 GB~31 tok/s- Qwen3.6 35B-A3B36BQ8_0 · 39 GB~103 tok/s
- Qwen2.5 32B33BQ8_0 · 37.6 GB~9 tok/s
- DeepSeek R1 Distill Qwen 32B33BQ8_0 · 37.6 GB~9 tok/s
- Qwen3 32B33BQ8_0 · 37.6 GB~9 tok/s
- QwQ 32B33BQ8_0 · 37.6 GB~9 tok/s
- Qwen2.5 Coder 32B33BQ8_0 · 37.6 GB~9 tok/s
Nemotron 3 Nano 30B-A3B32BQ8_0 · 34.2 GB~88 tok/s
Gemma 4 31B31BQ8_0 · 35 GB~10 tok/s
GLM-4.7 Flash 30B-A3B31BQ8_0 · 34.2 GB~103 tok/s- Qwen3 30B-A3B31BQ8_0 · 33.8 GB~93 tok/s
- Qwen3 Coder 30B-A3B31BQ8_0 · 33.8 GB~93 tok/s
- Qwen3.8 27B28BQ8_0 · 30.7 GB~11 tok/s
- Qwen3.6 27B28BQ8_0 · 30.7 GB~11 tok/s
- Gemma 3 27B27BQ8_0 · 30.8 GB~11 tok/s
- Gemma 2 27B27BQ8_0 · 31.8 GB~11 tok/s
- Gemma 4 26B-A4B26BQ8_0 · 28.3 GB~81 tok/s
- Mistral Small 3.2 24B24BQ8_0 · 27.4 GB~13 tok/s
- Mistral Small 24B24BQ8_0 · 27 GB~13 tok/s
- Devstral Small 24B24BQ8_0 · 27 GB~13 tok/s
- Magistral Small 24B24BQ8_0 · 27 GB~13 tok/s
gpt-oss 20B21BFP16 · 42.6 GB~45 tok/s- Qwen3 14B15BFP16 · 31.5 GB~11 tok/s
- Qwen2.5 Coder 14B15BFP16 · 31.8 GB~11 tok/s
- DeepSeek R1 Distill Qwen 14B15BFP16 · 31.8 GB~11 tok/s
- Qwen2.5 14B15BFP16 · 31.6 GB~11 tok/s
Phi-4 14B15BFP16 · 31.7 GB~11 tok/s- Phi-4 Reasoning 14B15BFP16 · 31.7 GB~11 tok/s
- Mistral Nemo 12B12BFP16 · 26.3 GB~13 tok/s
- Gemma 3 12B12BFP16 · 25.9 GB~13 tok/s
- Gemma 4 12B12BFP16 · 24.9 GB~14 tok/s
- Gemma 2 9B9.2BFP16 · 21.2 GB~18 tok/s
- Qwen3 8B8.2BFP16 · 18.2 GB~20 tok/s
- DeepSeek R1 0528 Qwen3 8B8.2BFP16 · 18.2 GB~20 tok/s
Granite 3.3 8B8.2BFP16 · 18.3 GB~20 tok/s- Llama 3.1 8B8BFP16 · 17.7 GB~20 tok/s
- Qwen2.5 7B7.6BFP16 · 16.3 GB~21 tok/s
- Qwen2.5 Coder 7B7.6BFP16 · 16.3 GB~21 tok/s
- DeepSeek R1 Distill Qwen 7B7.6BFP16 · 16.3 GB~21 tok/s
- Mistral 7B v0.37.3BFP16 · 16.2 GB~23 tok/s
- Gemma 3 4B4.3BFP16 · 9.5 GB~38 tok/s
- Qwen3 4B4BFP16 · 9.8 GB~41 tok/s
- Phi-4 Mini 3.8B3.8BFP16 · 9.4 GB~43 tok/s
- Phi-3.5 Mini 3.8B3.8BFP16 · 11.5 GB~43 tok/s
- Llama 3.2 3B3.2BFP16 · 8 GB~51 tok/s
- Qwen3 1.7B2BFP16 · 5.6 GB~81 tok/s
- Llama 3.2 1B1.2BFP16 · 3.3 GB~132 tok/s
- Gemma 3 1B1BFP16 · 2.6 GB~164 tok/s
- Qwen3 0.6B0.8BFP16 · 3 GB~218 tok/s
4 Fits with heavy quantization
NameMemory usedBest quantSpeed
12 Too big
NameMemory usedBest quantSpeed
Kimi K3.1 Pro 2.8T-A104B2800Bneeds 1698.3 GB48 GB free- Qwen3.8 2.4T-A95B2446Bneeds 1484.3 GB48 GB free
- DeepSeek V4 Pro 1.6T-A49B1600Bneeds 971.2 GB48 GB free
- Mistral Large 4 1T-A52B1000Bneeds 607.5 GB48 GB free
- DeepSeek V4.1 Flash763Bneeds 463.4 GB48 GB free
- GLM-5.3753Bneeds 457.8 GB48 GB free
- GLM-5 744B-A40B744Bneeds 452.4 GB48 GB free
MiniMax M3 428B-A23B427Bneeds 260.5 GB48 GB free- GLM-5.3 Flash 320B-A18B321Bneeds 195.3 GB48 GB free
- DeepSeek V4 Flash 284B-A13B284Bneeds 173 GB48 GB free
- Qwen3 235B-A22B235Bneeds 144.7 GB48 GB free
- Qwen3.8 Flash Next180Bneeds 109.9 GB48 GB free
Run it in the cloud
Need more than 64 GB? Rent a bigger GPU by the hour.
Rent a cloud GPU by the hour. No upfront cost, stop any time.