This model is 22 months old. A newer version of it has since shipped: Mistral Large 4 (2026-10-06). The details below are still accurate for this model. It just is not what we would recommend today.
Mistral Small 3
Mistral AI🇫🇷 France
Context window33K tokens
Input / 1M tokens$0.05
Output / 1M tokens$0.08
Pricing and context window via OpenRouter, 2026-11-21.
Version History
24b-instruct-2501major
Mistral Small 3 introduces a new 24B-parameter model optimized for inference speed, delivering 3x faster performance than 70B models while maintaining competitive accuracy at 81% MMLU.