Mistral: Mistral Small 3.1 24B

Mistral Small 3.1 24B from MistralAI is designed for handling both text and image inputs within a context length of up to 128,000 tokens. It excels in processing extensive data but does not support tools or reasoning tasks nor provide structured outputs. The model's benchmark score stands at 30.4 on EQBench as of August 2026, contributing to a blended score of 26.8 across one benchmark. This model is best suited for projects requiring text and image processing capabilities within its context limits, though it may be less ideal for more complex tasks that involve reasoning or structured data handling. Considering its pricing at $0.351 per million input tokens and $0.555 per million output tokens, users should weigh these costs against the model's performance in their specific needs, especially given its strong standing on EQBench.