Terminally online Mistral Vibe.

Mistral AI announced the release of Mistral Vibe 2.0, described as a major upgrade to its terminal-native coding agent. The release is powered by the Devstral 2 model family.

According to the announcement, Vibe 2.0 adds custom subagents for targeted tasks such as deploy scripts, pull request reviews, and test generation, invoked on demand. It also introduces multi-choice clarifications, where Vibe prompts with options when intent is ambiguous instead of guessing, and slash-command skills that load preconfigured workflows for tasks like deploying, linting, or generating documentation. Unified agent modes let users configure custom modes combining tools, permissions, and behaviors. Bug fixes and improvements in the Vibe CLI now ship continuously, with no manual updates required.

Mistral said Vibe is available on Le Chat Pro and Team plans, with pay-as-you-go credits at API rates for usage beyond plan limits, or users can bring their own API key. Students get 50% off. Devstral 2 moves to paid API access, with free API usage remaining on the Experiment plan in Mistral Studio. Devstral 2 pricing is listed at $0.40 per million input tokens and $2.00 per million output tokens; Devstral 2 Small is listed at $0.10 per million input tokens and $0.30 per million output tokens.

Mistral also detailed enterprise add-ons: fine-tuning on internal languages and DSLs, reinforcement learning with a customer's own environment, and end-to-end code modernization. It said these solutions are already delivered for organizations in finance, defense, and infrastructure. Installation instructions for the Vibe CLI were provided, along with a sign-up step and documentation. Mistral added that it is hiring.