Mistral AI Devstral Models Release

Mistral AI, in collaboration with All Hands AI, has introduced Devstral Medium and an upgraded Devstral Small 1.1. These models are designed to enhance agentic coding capabilities, emphasizing generalization across various prompts and agentic scaffolds to improve autonomous software development.

Devstral Small 1.1: Open-Source Agentic Coding

Devstral Small 1.1 is a 24B parameter model released under the Apache 2.0 license, establishing a new state-of-the-art for open models without test-time scaling. It is designed for local deployment and community customization.

Performance and Benchmarks

Devstral Small 1.1 achieves a score of 53.6% on SWE-bench Verified. This performance makes it the leading open model for code agents in its category.

Versatility and Integration

The model demonstrates improved generalization to coding environments and prompts. It is specifically optimized for use with OpenHands and supports both XML formats and Mistral function calling, allowing it to be integrated into a wide range of agentic scaffolds.

Devstral Medium: High-Performance Enterprise Coding

Devstral Medium is a high-performance model available via API, designed to balance cost and performance for businesses and developers. It surpasses Gemini 2.5 Pro and GPT 4.1 in performance while costing a quarter of the price.

Performance and Deployment Options

Devstral Medium achieves a score of 61.6% on SWE-bench Verified. For enterprise users, the model is available through the public API, can be deployed on private infrastructure for enhanced data privacy, and supports custom finetuning via the finetuning API to tailor performance to specific requirements.

Model Availability and Pricing

Both models are accessible via the Mistral AI API with the following pricing structures:

  • devstral-small-2507: Priced at $0.1 per million input tokens and $0.3 per million output tokens (matching Mistral Small 3.1).
  • devstral-medium-2507: Priced at $0.4 per million input tokens and $2 per million output tokens (matching Mistral Medium 3).

Devstral Small 1.1 is available for download via its model card on Hugging Face under the Apache 2.0 license. Devstral Medium is available for enterprise customers via Mistral Code and the finetuning API.

Sources

Related

  • Dispatch
    Mistral AI Devstral Release

    Mistral AI has released Devstral, an agentic LLM for software engineering that outperforms open-source models on the SWE-Bench Verified benchmark and is available under the Apache 2.0 license.

  • Dispatch
    Mistral AI Devstral 2 and Mistral Vibe CLI Release

    Mistral AI has released Devstral 2 (123B) and Devstral Small 2 (24B), open-weight coding models that achieve state-of-the-art performance on SWE-bench Verified with high cost-efficiency, alongside the open-source Mistral Vibe CLI for autonomous code automation.

  • Dispatch
    Mistral AI Codestral 25.08 and Enterprise Coding Stack Release

    Mistral AI has released Codestral 25.08 and a comprehensive enterprise coding stack featuring integrated completion, semantic search, and agentic workflows with flexible deployment options.

  • Dispatch
    Mistral Small 3.1 Release Notes

    Mistral AI has released Mistral Small 3.1, an Apache 2.0 licensed model featuring multimodal understanding, a 128k token context window, and performance that surpasses GPT-4o Mini and Gemma 3.

  • Dispatch
    Mistral Medium 3 Release Notes

    Mistral AI has released Mistral Medium 3, a frontier-class model that balances state-of-the-art performance with 8x lower costs and flexible enterprise deployment options.