Skip to main content

Devstral-Small-2-24B-Instruct-2512

Description​

“Devstral-Small-2-24B-Instruct-2512” is engineered for agentic coding workflows and tool integration. It dynamically routes tasks across specialized skills to maximize efficiency and context awareness and excels at orchestrating complex, multi-step programming and automation pipelines.

It supports and is suitable for:

  • Text generation within a chat completion (text to text)
  • Tool-calling for agentic workflows
  • Image understanding (vision)

The following limitations apply:

  • Maximum context length: 262,144 tokens
  • Maximum images per context: 4
  • Images must be submitted as Base64-encoded data URLs (no remote URLs)
  • This model does not support thinking / reasoning

It is recommended to operate the model with the following parameters:

  • temperature=0.7
  • top_p=0.8
  • top_k=20
  • repetition_penalty=1.05

For vision tasks, lower temperature (temperature=0.1) is recommended when accurate image description or text extraction is required. Always resize images to a maximum of 1024 px on the longest edge before encoding - large images significantly increase time to first token (TTFT). See the Python examples or JavaScript examples for a ready-to-use helper.

Terms of use and licensing​

The general terms of use apply. The model is provided under the Apache 2.0 License, and reuse of the generated content is not subject to any additional restrictions.