🔌
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 48GB run CodeT5+?
Yes — here's how fast and up to what context, with real measured numbers.
CodeT5+ on the Apple M3 Max 48GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 37k | 8.5-10.3 t/s | ~0.2-0.3 (slow) |
| Q8 | yes | 98k | 17-20.5 t/s | ~0.4-0.5 (slow) |
| Q4 | yes | 125k | 30.4-36.6 t/s | ~0.8-0.9 (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 48GB run CodeT5+?▶
Yes. At Q4 it generates ~30.4-36.6 tok/s and fits up to 125k context; beyond that it spills to SSD swap and slows to ~0.8-0.9 tok/s.
How much context fits?▶
At Q4, up to about 125k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.