Mistral: Mistral Small 3.2 24B

Mistral Small 3.2 24B is a multimodal model from Mistral AI that accepts both image and text inputs, supports tool use, and offers a 128,000-token context window with up to 16,384 tokens of output per completion. It does not include a dedicated reasoning mode, and structured output support is unconfirmed. The combination of vision capability and tool use makes it broadly applicable to agentic and document-processing workflows without requiring a larger, more expensive model. At $0.075 per million input tokens and $0.20 per million output tokens, it sits in the budget-to-midrange tier, making it worth considering for teams running high-volume inference who need image understanding alongside text. Benchmark coverage is thin, with a blended score of 59.6 drawn from a single benchmark, so performance claims should be treated as provisional rather than well-established. Teams that need wider benchmark validation before committing should treat it as a candidate to test rather than a settled choice.