// RUN โ€” FITS

Can the Apple M1 Ultra 128GB run Gemma 3 12B?

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

Gemma 3 12B on the Apple M1 Ultra 128GB

QuantFits?Max contextSpeedBeyond context
FP16 yes 547k 22.7-27.3 t/s ~0.3-0.3 (slow)
Q8 yes 621k 45.3-54.7 t/s ~0.6-0.7 (slow)
Q4 yes 653k 81-97.6 t/s ~1-1.2 (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 Ultra 128GB run Gemma 3 12B?โ–ถ

Yes. At Q4 it generates ~81-97.6 tok/s and fits up to 653k context; beyond that it spills to SSD swap and slows to ~1-1.2 tok/s.

How much context fits?โ–ถ

At Q4, up to about 653k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.