magistral-small-2506

Mistral AI's 24B-parameter, open-source LLM (Magistral-Small-2506) is purpose-built for transparent, multi-step reasoning across diverse enterprise use cases.

Magistral-Small-2506 is Mistral AI’s 24B-parameter, open-source Large Language Model, released under the permissive Apache 2.0 license. This model is fine-tuned for transparent, multi-step reasoning, making it ideal for complex enterprise challenges: structured calculations, programmatic logic, and financial forecasting. It demonstrates strong performance, scoring 70.7% on key reasoning benchmarks, and supports a 128k context window (40k recommended). Deployment is efficient: the quantized GGUF version fits locally on a single NVIDIA RTX 4090 or a 32GB MacBook, enabling rapid, on-premise development and fine-tuning with frameworks like vLLM and Ollama.

https://huggingface.co/mistralai/Magistral-Small-2506