§01·spec · /gpus
Apple M2 Max
appleapple series64GB unified
26 open-weights AI models run on the Apple M2 Max — see which fit, how fast they go, and the VRAM each needs.
Apple Silicon uses unified memory shared between CPU and GPU. Of this machine’s 64 GB the GPU addresses 48.0 GiB by default — exactly three quarters — and the limit is raisable via Metal’s wired-memory setting.
§02·models that run on this GPU
26 totalLLM · 12
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| Devstral Small 2 (24B) | 64GB | ✓ | recipe | check ↗ | |
| gpt-oss 20B | 13GB | ✓ | recipe | check ↗ | |
| KAT-Coder V2.5 Dev | 48GB | ✓ | recipe | check ↗ | |
| Laguna XS 2.1 | 24GB | ✓ | recipe | check ↗ | |
| Llama 3.1 8B | 5GB | ✓ | recipe | check ↗ | |
| Llama 3.3 70B | 40GB | ✓ | recipe | check ↗ | |
| Mistral Small 3.2 24B | 64GB | ✓ | recipe | check ↗ | |
| Nanbeige4.2 3B | 16GB | ✓ | recipe | check ↗ | |
| North Mini Code 1.0 | 48GB | ✓ | recipe | check ↗ | |
| Ornith 1.0 35B | 48GB | ✓ | recipe | check ↗ | |
| Qwen3 32B | 19GB | ✓ | recipe | check ↗ | |
| Qwen3-Next 80B-A3B | 64GB | ✓ | recipe | check ↗ |
- Best speed
- Min VRAM
- 64GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 13GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 48GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 24GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 5GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 40GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 64GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 16GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 48GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 48GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 19GB
- Evidence
- recipe
- Best speed
- Min VRAM
- 64GB
- Evidence
- recipe
Multimodal · 6
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| Agents-A1 4B | 14GB | ✓ | recipe | check ↗ | |
| Fara1.5-27B | 64GB | ✓ | recipe | check ↗ | |
| Fara1.5-4B | 24GB | ✓ | recipe | check ↗ | |
| Fara1.5-9B | 32GB | ✓ | recipe | check ↗ | |
| Gemma 4 E4B-IT | 5GB | ✓ | recipe | check ↗ | |
| Qwen3.8 27B | 48GB | ✓ | recipe | check ↗ |
Image · 3
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| Krea 2 | 24GB | ✓ | recipe | check ↗ | |
| Qwen-Image | 23GB | ✓ | recipe | check ↗ | |
| Z-Image Turbo | 17GB | ✓ | recipe | check ↗ |
Video · 2
3D · 1
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| TRELLIS.2-4B | 17GB | ✓ | recipe | check ↗ |
§03·tested recipes
showing 6 of 26- multimodalintermediate48GB+recipe
Qwen3.8-27B on Apple M2 Max: 8-bit MLX Vision-Language with MTP Speculative Decoding
- llmintermediate16GB+recipe
Nanbeige4.2-3B on Apple M2 Max: 128K-context agentic LLM via llama.cpp Metal
- videoadvanced64GB+recipe
MiniMax H3 on Apple M2 Max: native MLX video with synchronised audio
- multimodaladvanced32GB+recipe
Fara1.5-9B on Apple M2 Max: Browser Computer-Use Agent on llama.cpp Metal
- multimodaladvanced24GB+recipe
Fara1.5-4B on Apple M2 Max: Browser Computer-Use Agent on llama.cpp Metal
- multimodaladvanced64GB+recipe
Fara1.5-27B on Apple M2 Max: Browser Computer-Use Agent at the Full 262K Context