🔌
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 M1 Max 32GB run XLM-RoBERTa?
Yes — here's how fast and up to what context, with real measured numbers.
XLM-RoBERTa on the Apple M1 Max 32GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 400k | 247.3-298.2 t/s | ~6.2-7.5 (slow) |
| Q8 | yes | 408k | 494.5-596.4 t/s | ~12.4-14.9 (slow) |
| Q4 | yes | 412k | 883.1-1064.9 t/s | ~22.1-26.6 (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 M1 Max 32GB run XLM-RoBERTa?▶
Yes. At Q4 it generates ~883.1-1064.9 tok/s and fits up to 412k context; beyond that it spills to SSD swap and slows to ~22.1-26.6 tok/s.
How much context fits?▶
At Q4, up to about 412k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.