What LLMs Can the Apple M1 Max 64GB Run?

The Apple M1 Max 64GB has 64 GB of unified memory (about 48.0 GB usable by the GPU) and 400 GB/s of bandwidth. It runs 46 of 58 models fully at Q4_K_M with 8K context, up to Qwen3-Next 80B-A3B, and generates about 47 tokens/s on an 8B model.

Apple · Apple Silicon · Released 2021-10 · Updated October 14, 2026

46 of 58 models fit

Apple M1 Max 64GB (48.0 GB GPU-usable) at Q4_K_M with 8K context. Largest that fits fully: Qwen3-Next 80B-A3B.

ModelNeedsStatusSpeedBest quant that fits
Qwen 2.5 72B44.5 GBFits~5.7 tok/sQ4_K_M
Llama 3.3 70B43.2 GBFits~5.9 tok/sQ4_K_M
DeepSeek R1 Distill Llama 70B43.2 GBFits~5.9 tok/sQ4_K_M
Mixtral 8x7B28.0 GBFits~28 tok/sQ6_K
Qwen 2.5 32B21.4 GBFits~12 tok/sQ8_0
Qwen 2.5 Coder 32B21.4 GBFits~12 tok/sQ8_0
DeepSeek R1 Distill Qwen 32B21.4 GBFits~12 tok/sQ8_0
Qwen 3 32B21.4 GBFits~12 tok/sQ8_0
Qwen 3.5 35B-A3B21.0 GBFits~86 tok/sQ8_0
Qwen 3.6 35B-A3B21.0 GBFits~86 tok/sQ8_0
Gemma 4 31B19.8 GBFits~13 tok/sQ8_0
Qwen 3 30B-A3B18.8 GBFits~71 tok/sQ8_0
Qwen3-Coder 30B-A3B18.8 GBFits~71 tok/sQ8_0
Gemma 3 27B17.6 GBFits~15 tok/sQ8_0
Qwen 3.5 27B16.9 GBFits~15 tok/sQ8_0
Qwen 3.6 27B16.9 GBFits~15 tok/sQ8_0
Qwen 3.8 27B16.9 GBFits~15 tok/sQ8_0
Mistral Small 3.1 24B15.6 GBFits~17 tok/sQ8_0
Gemma 4 26B-A4B15.6 GBFits~77 tok/sQ8_0
gpt-oss-20b12.9 GBFits~89 tok/sMXFP4 (official)
Qwen 2.5 14B10.7 GBFits~26 tok/sFP16 / BF16
DeepSeek R1 Distill Qwen 14B10.7 GBFits~26 tok/sFP16 / BF16
Phi-4 14B10.6 GBFits~26 tok/sFP16 / BF16
Qwen 3 14B10.5 GBFits~27 tok/sFP16 / BF16
Ministral 3 14B9.7 GBFits~29 tok/sFP16 / BF16
Mistral Nemo 12B9.0 GBFits~32 tok/sFP16 / BF16
Gemma 3 12B8.8 GBFits~32 tok/sFP16 / BF16
Ministral 3 8B6.7 GBFits~45 tok/sFP16 / BF16
Qwen 3 8B6.6 GBFits~46 tok/sFP16 / BF16
Qwen 3.5 9B6.4 GBFits~46 tok/sFP16 / BF16
Llama 3.1 8B6.3 GBFits~47 tok/sFP16 / BF16
DeepSeek R1 Distill Llama 8B6.3 GBFits~47 tok/sFP16 / BF16
Mistral 7B v0.35.7 GBFits~52 tok/sFP16 / BF16
Gemma 4 E4B5.7 GBFits~52 tok/sFP16 / BF16
Qwen 2.5 7B5.6 GBFits~53 tok/sFP16 / BF16
DeepSeek R1 Distill Qwen 7B5.6 GBFits~53 tok/sFP16 / BF16
Qwen 3 4B4.2 GBFits~83 tok/sFP16 / BF16
Phi-4 Mini4.1 GBFits~88 tok/sFP16 / BF16
Gemma 4 E2B4.0 GBFits~81 tok/sFP16 / BF16
Gemma 3 4B3.7 GBFits~90 tok/sFP16 / BF16
Qwen 3.5 4B3.7 GBFits~92 tok/sFP16 / BF16
Ministral 3 3B3.5 GBFits~101 tok/sFP16 / BF16
Llama 3.2 3B3.5 GBFits~104 tok/sFP16 / BF16
Llama 3.2 1B1.7 GBFits~280 tok/sFP16 / BF16
Gemma 3 1B1.6 GBFits~370 tok/sFP16 / BF16
Qwen3-Next 80B-A3B46.3 GBTight fit~80 tok/sIQ4_XS
Kimi K2.6566 GBToo small––
DeepSeek R1 671B380 GBToo small––
Llama 3.1 405B234 GBToo small––
Qwen 3.5 397B-A17B226 GBToo small––
DeepSeek V4 Flash161 GBToo small––
Qwen 3 235B-A22B135 GBToo small––
MiniMax-M2133 GBToo small––
Qwen 3.5 122B-A10B70.2 GBToo small–Q2_K
Mistral Small 4 119B68.2 GBToo small–Q2_K
gpt-oss-120b67.2 GBToo small–Q2_K
Llama 4 Scout64.0 GBToo small–Q2_K
GLM-4.5-Air62.1 GBToo small–Q2_K

Apple M1 Max 64GB for local AI

A popular used choice for local LLMs. About 48 GB is GPU-usable by default, enough for 70B models at Q4_K_M with a short context, at roughly three quarters of an M4 Max's speed.

Among the 6 devices we track with about 48 GB of usable memory, the Apple M1 Max 64GB ranks 5 of 6 for text-generation speed, because generation speed follows memory bandwidth. Others in the same class: RTX 6000 Ada 48GB, L40S 48GB, RTX A6000 48GB, Apple M4 Max 64GB. They all hold the same models; the faster ones simply answer sooner.

Our picks for the Apple M1 Max 64GB

The model we would run first for each job, chosen from our recommendations by what fits fully with an 8K context. "Best quant" is the highest quality that fits; "max context" is at Q4_K_M or the native format.

TaskModelBest quantMax contextSpeed
Best all-rounderLlama 3.3 70BQ4_K_M14K~5.9 tok/s
CodingQwen 3.6 27BQ8_0256K~15 tok/s
Fastest repliesQwen 3.6 35B-A3BQ8_0256K~86 tok/s
Long documents (64K+)Qwen 3.8 27BQ8_0256K~15 tok/s
Images in, text outGemma 4 31BQ8_0256K~13 tok/s
Image generationFLUX.2 [dev]FP8––
SpecValue
Memory64 GB unified (48.0 GB GPU-usable by default)
Memory bandwidth400 GB/s
ArchitectureApple Silicon
Released2021-10
Llama 3.1 8B Q4_K_M speed~47 tok/s
Largest model that fitsQwen3-Next 80B-A3B

Image and video models on the Apple M1 Max 64GB

Peak VRAM with model swapping (ComfyUI default on smaller cards).

ModelFP8FP8 fitGGUF Q4Q4 fit
Stable Diffusion 1.51.8 GBFits1.7 GBFits
SDXL 1.04.1 GBFits4.1 GBFits
Stable Diffusion 3.5 Medium6.4 GBFits6.4 GBFits
Stable Diffusion 3.5 Large8.6 GBFits6.4 GBFits
FLUX.1 [dev]12.1 GBFits7.3 GBFits
FLUX.1 [schnell]12.1 GBFits7.3 GBFits
Qwen-Image20.4 GBFits12.3 GBFits
Wan 2.2 A14B (T2V / I2V)18.8 GBFits13.2 GBFits
Wan 2.2 TI2V 5B8.0 GBFits6.5 GBFits
HunyuanVideo18.3 GBFits13.1 GBFits
FLUX.1 Kontext [dev]12.3 GBFits7.5 GBFits
FLUX.2 [dev]31.0 GBFits23.6 GBFits
Z-Image Turbo6.8 GBFits4.9 GBFits
Qwen-Image-Edit 250920.5 GBFits12.4 GBFits
LTX-219.2 GBFits12.6 GBFits

Upgrade path

The next step up is the Apple M3 Max 128GB with 128 GB.

Apple M1 Max 64GB for LLMs: common questions

Can the Apple M1 Max 64GB run Llama 3.3 70B?

Yes. Llama 3.3 70B needs about 43.2 GB at Q4_K_M with 8K context, and the best quality that fits is Q4_K_M.

Can the Apple M1 Max 64GB run Qwen 3 32B?

Yes. Qwen 3 32B needs about 21.4 GB at Q4_K_M with 8K context, and the best quality that fits is Q8_0.

Can the Apple M1 Max 64GB run Qwen 3 14B?

Yes. Qwen 3 14B needs about 10.5 GB at Q4_K_M with 8K context, and the best quality that fits is FP16 / BF16.

How fast is the Apple M1 Max 64GB for LLMs?

About 47 tokens per second for an 8B model at Q4_K_M, based on its 400 GB/s memory bandwidth. Larger models scale down roughly in proportion to their size.

How much of the 64 GB can the GPU use?

About 48.0 GB by default. You can raise it with sudo sysctl iogpu.wired_limit_mb=<MB> (resets on reboot), leaving at least 8 GB for macOS.

Related