Key Info

Mistral has launched the Responses API in public preview, designed for building agents, and is now serving two additional fast, low-cost models on Mistral AI Cloud Inference in the EU and US.

Highlights

  • Each Responses API call returns the model's answer, its reasoning and its tool calls as separate items, making step-by-step handling easier than with Chat Completions.
  • The API works with all chat models hosted by Mistral, including Large 4.
  • DeepSeek V4.1 Flash is in public preview, positioned for long-context coding and agentic tasks.
  • GLM 5.3 Flash is generally available, positioned for subagents, tool calls and coding tasks.