Mistral Medium 3.5 This model is a dense one of 128B parameters, including vision: able to perform instruct, reasoning and agentic tasks. architecture is the same as Mistral 3 with yarn rope. the way to trigger reasoning is the same as Mistral Small 4 which is to pass reasoning effort="high" for reasoning and reasoning effort="none" for instruct. context length of 256k Expect: gap of performance in comparison with Mistral Small 4 for instruct reasoning and agentic tasks. Can replace Devstral 2 for agentic. a known limitation is still a multimodal performance subpar with the competition. Recommended Settings Reasoning Effort : 'none' → Do not use reasoning 'high' → Use reasoning (recommended for complex prompts) Use reasoning effort="high" for complex tasks Temperature : 0.7 for reasoning effort="high" . Temp between 0.0 and 0.7 for reasoning effort="none" depending on task. Use Cases Mistral Small 4 is designed for general chat assistants, coding, agentic tasks, and reasoning tasks (with reasoning mode toggled). Its multimodal capabilities also enable document and image understanding for data extraction and analysis. Its capabilities are ideal for: Developers interested in coding a…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy