🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN ON APPLE
Run Mage-Flow on Apple hardware
22 options with enough VRAM (FP16), ranked by headroom.
| Hardware | VRAM | Fits FP16? | Headroom | ~Price |
|---|---|---|---|---|
| Apple M2 Ultra 192GB → | 192 GB | yes | ~163.3 GB | ~$6,600 |
| Apple M1 Ultra 128GB → | 128 GB | yes | ~105.7 GB | ~$4,800 |
| Apple M4 Max 128GB → | 128 GB | yes | ~105.7 GB | ~$4,900 |
| Apple M3 Max 128GB → | 128 GB | yes | ~105.7 GB | ~$4,700 |
| GMKtec EVO-X3 → | 128 GB | yes | ~105.7 GB | — |
| Apple M2 Max 96GB → | 96 GB | yes | ~76.9 GB | ~$3,300 |
| Apple M2 Ultra 64GB → | 64 GB | yes | ~48.1 GB | ~$4,000 |
| Apple M1 Ultra 64GB → | 64 GB | yes | ~48.1 GB | ~$3,700 |
| Apple M4 Max 64GB → | 64 GB | yes | ~48.1 GB | ~$3,900 |
| Apple M3 Max 64GB → | 64 GB | yes | ~48.1 GB | ~$3,700 |
| Apple M1 Max 64GB → | 64 GB | yes | ~48.1 GB | ~$2,600 |
| Apple M2 Max 64GB → | 64 GB | yes | ~48.1 GB | ~$2,800 |
| Apple M4 Max 48GB → | 48 GB | yes | ~33.7 GB | ~$3,300 |
| Apple M3 Max 48GB → | 48 GB | yes | ~33.7 GB | ~$3,200 |
| Apple M4 Max 36GB → | 36 GB | yes | ~22.9 GB | ~$3,000 |
| Apple M3 Max 36GB → | 36 GB | yes | ~22.9 GB | ~$2,900 |
| Apple M1 Max 32GB → | 32 GB | yes | ~19.3 GB | ~$2,000 |
| Apple M2 Max 32GB → | 32 GB | yes | ~19.3 GB | ~$2,200 |
| Jetson Thor T3000 → | 32 GB | yes | ~19.3 GB | — |
| NVIDIA Jetson T3000 → | 32 GB | yes | ~19.3 GB | — |
| NVIDIA Jetson T2000 → | 16 GB | yes | ~4.9 GB | — |
| Jetson Thor T2000 → | 16 GB | yes | ~4.9 GB | — |
VRAM fit at FP16 (lower precision needs less). We don't estimate generation speed for image/video/audio — it depends on tool, resolution and steps.
FAQ
What Apple hardware can run Mage-Flow?▶
On GenAIList, 22 Apple GPUs have enough VRAM to hold Mage-Flow at FP16 — the most headroom is on the Apple M2 Ultra 192GB (192 GB).
How fast is Mage-Flow on Apple hardware?▶
We don't estimate generation speed for image/video/audio models — it depends on the tool, resolution and step count. This page ranks by which hardware fits the model in memory.