Mistral Small 3.1

Mistral Small 3.1 (2503) builds upon Mistral Small 3 by adding state-of-the-art vision understanding and enhancing long context capabilities up to 128K tokens without compromising text performance. With 24 billion parameters, this model achieves top-tier capabilities in both text and vision tasks while remaining efficient enough to run on a single GPU.

The model outperforms comparable models like Gemma 3 and GPT-4o Mini across a range of benchmarks, while delivering inference speeds of 150 tokens per second. It supports function calling, structured outputs, and JSON mode — making it suitable for agentic workflows and tool-use scenarios.

Mistral Small 3.1 is released under the Apache 2.0 license, making it fully open-source and available for commercial use without restrictions. On CallMissed, it runs on the CallMissed gateway, making it available on the free tier with no additional cost beyond credits.

Key improvements over Mistral Small 3 include multimodal vision understanding (the model can process images alongside text), extended context from 32K to 128K tokens, and improved performance on long-document comprehension tasks. The model is optimized for efficient local inference, supporting use cases such as conversational agents, function calling, long-document comprehension, and privacy-sensitive deployments.