// RUN — FITS

Can the Apple M3 Max 36GB run ULM-FiT?

Yes — here's how fast and up to what context, with real measured numbers.

ULM-FiT on the Apple M3 Max 36GB

QuantFits?Max contextSpeedBeyond context
FP16 yes 458k 231.3-278.9 t/s ~7.7-9.3 (slow)
Q8 yes 465k 462.6-557.8 t/s ~15.4-18.6 (slow)
Q4 yes 468k 826-996.1 t/s ~27.5-33.2 (slow)

Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.

→ All models on the Apple M3 Max 36GB → ULM-FiT specs → Other hardware

FAQ

Can the Apple M3 Max 36GB run ULM-FiT?▶

Yes. At Q4 it generates ~826-996.1 tok/s and fits up to 468k context; beyond that it spills to SSD swap and slows to ~27.5-33.2 tok/s.

How much context fits?▶

At Q4, up to about 468k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.