Decode · tok/s
70
Prefill
—
tok/100W
20
Here is what 1 used 3090 does with it on Strata: IQ3_S: 70 tok/s, and still 66 tok/s with 261K of context IQ2_XS: 87 tok/s prompts are read at about 1,900 tok/s
First-person single 3090. Rank is the quoted 70. 66 at 261K and prefill ~1,900 stay in the note. IQ2_XS 87 is a separate pair.
