Mistral: Devstral Small 1.1

Devstral Small 1.1 is a powerful 24B parameter open-weight language model specifically designed for software engineering agents. Developed by Mistral AI in collaboration with All Hands AI, it's finetuned from Mistral Small 3.1 and released under the permissive Apache 2.0 license. This model boasts an impressive 128k token context window and supports both Mistral-style function calling and XML output formats, making it highly versatile for complex development tasks. Optimized for agentic coding workflows, Devstral Small 1.1 excels at tasks such as codebase exploration, multi-file edits, and seamless integration into autonomous development agents like OpenHands and Cline. It achieves an outstanding 53.6% on SWE-Bench Verified, surpassing all other open models on this benchmark, while remaining lightweight enough to run on a single 4090 GPU or Apple silicon machine. The model uses a Tekken tokenizer with a 131k vocabulary and is deployable via vLLM, Transformers, Ollama, LM Studio, and other OpenAI-compatible runtimes. Access Devstral Small 1.1 for free on Multi AI, with pricing at $0.10/0.30 per 1M tokens (input/output) for advanced usage. It supports functions, code generation, and streaming, with a max output of 4K tokens.