Mistral Small 3 Release: 24B Open Model, Speed, MMLU
Mistral released Small 3 under Apache 2.0, reporting 81% MMLU accuracy at 150 tokens per second and more than triple the speed of larger models on comparable hardware.
Mistral released Small 3 under Apache 2.0, reporting 81% MMLU accuracy at 150 tokens per second and more than triple the speed of larger models on comparable hardware.