// RUN โ€” FITS

Can the Jetson Thor T3000 run ESM1b?

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

ESM1b on the Jetson Thor T3000

QuantFits?Max contextSpeedBeyond context
FP16 yes 397k 152.7-182 t/s ~5.6-6.7 (slow)
Q8 yes 407k 305.5-364.1 t/s ~11.2-13.3 (slow)
Q4 yes 411k 545.5-650.1 t/s ~20-23.8 (slow)

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

FAQ

Can the Jetson Thor T3000 run ESM1b?โ–ถ

Yes. At Q4 it generates ~545.5-650.1 tok/s and fits up to 411k context; beyond that it spills to SSD swap and slows to ~20-23.8 tok/s.

How much context fits?โ–ถ

At Q4, up to about 411k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.