🔌
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 M3 Max 36GB run SantaCoder?
Yes — here's how fast and up to what context, with real measured numbers.
SantaCoder on the Apple M3 Max 36GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 438k | 92.7-111.8 t/s | ~3.1-3.7 (slow) |
| Q8 | yes | 455k | 185.5-223.6 t/s | ~6.2-7.5 (slow) |
| Q4 | yes | 462k | 331.2-399.4 t/s | ~11-13.3 (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 M3 Max 36GB run SantaCoder?▶
Yes. At Q4 it generates ~331.2-399.4 tok/s and fits up to 462k context; beyond that it spills to SSD swap and slows to ~11-13.3 tok/s.
How much context fits?▶
At Q4, up to about 462k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.