🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the Apple M4 Max 36GB run MT-DNN?
Yes — here's how fast and up to what context, with real measured numbers.
MT-DNN on the Apple M4 Max 36GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 461k | 562.5-678.4 t/s | ~10.3-12.4 (slow) |
| Q8 | yes | 466k | 1125.1-1356.7 t/s | ~20.6-24.8 (slow) |
| Q4 | yes | 469k | 2009.1-2422.7 t/s | ~36.8-44.4 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the Apple M4 Max 36GB run MT-DNN?▶
Yes. At Q4 it generates ~2009.1-2422.7 tok/s and fits up to 469k context; beyond that it spills to SSD swap and slows to ~36.8-44.4 tok/s.
How much context fits?▶
At Q4, up to about 469k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.