§01·spec · /gpus
Apple M4 Max
appleapple series48GB unified
13 open-weights AI models run on the Apple M4 Max — see which fit, how fast they go, and the VRAM each needs.
Apple Silicon uses unified memory shared between CPU and GPU. By default the GPU can address about 75% of it; the limit is raisable via Metal’s wired-memory setting.
§02·models that run on this GPU
13 totalLLM · 5
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| gpt-oss 20B | 13GB | ✓ | recipe | check ↗ | |
| Llama 3.1 8B | 5GB | ✓ | recipe | check ↗ | |
| Llama 3.3 70B | 40GB | ✓ | recipe | check ↗ | |
| Ornith 1.0 35B | 48GB | ✓ | recipe | check ↗ | |
| Qwen3 32B | 19GB | ✓ | recipe | check ↗ |
Multimodal · 1
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| Gemma 4 E4B-IT | 5GB | ✓ | 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 ↗ |
3D · 1
| Model | Best speed | Min VRAM | Works | Evidence | |
|---|---|---|---|---|---|
| TRELLIS.2-4B | 17GB | ✓ | recipe | check ↗ |
§03·tested recipes
showing 6 of 13- llmadvanced48GB+recipe
Ornith 1.0 35B on Apple M4 Max: Local Agentic Coding via llama.cpp Metal + OpenHands (48GB Unified Memory)
- imageintermediate24GB+recipe
Krea 2 Turbo on Apple M4 Max: 8-Step Text-to-Image in Unified Memory via ComfyUI (MPS)
- 3dadvanced17GB+recipe
TRELLIS.2-4B on Apple M4 Max: image-to-3D in unified memory via the community Metal port
- videoadvanced34GB+recipe
LTX-2.3 on Apple M4 Max: experimental 22B audio-video in unified memory via Draw Things (Metal) or MLX
- ttsbeginner2GB+recipe
VoxCPM-0.5B on Apple M4 Max: Zero-Shot Voice Cloning TTS in Unified Memory (MPS)
- ttsbeginner1GB+recipe
Kokoro TTS on Apple M4 Max: 82M Text-to-Speech, 54 Voices, Native MLX-Audio