2 ms·Mistral is nowhere near GPT-3.5 according to corrected benchmarks3 points by behnamoh 3y agosp332 3y agoThese benchmarks are for the older 7B model, now known as "Tiny". It does not include the 8x7B "Small" model, or the API-only "Medium".