title: "Upgrading agentic coding capabilities with the new Devstral models | Mistral AI"

Mistral AI has introduced Devstral Medium and an upgraded Devstral Small 1.1, two new models focused on enhancing agentic coding capabilities. The open-source Devstral Small 1.1 sets a new state-of-the-art for open models on the SWE-Bench Verified benchmark with a score of 53.6%. Devstral Medium, available via API, scores even higher at 61.6% on the same benchmark, outperforming competitors like Gemini 2.5 Pro and GPT 4.1 at a significantly lower cost. These models are designed for versatility, supporting various function calling formats and agentic frameworks to improve autonomous software development.