Mistral AI released Vibe 2.0, a major upgrade to its terminal-native coding agent that introduces custom subagents, multi-choice clarifications, and slash-command skills powered by the Devstral 2 model family.

The new version allows developers to build specialized agents for targeted tasks like deploy scripts, pull request reviews, and test generation. These custom subagents can be invoked on demand within the terminal environment.

Key features and improvements

Vibe 2.0 adds multi-choice clarifications that prompt users with options when intent is ambiguous, rather than making assumptions. The system now supports slash-command skills that load preconfigured workflows for common development tasks including deploying, linting, and generating documentation.

The upgrade introduces unified agent modes that combine tools, permissions, and behaviors. Developers can switch contexts without changing tools, and the CLI now ships continuous bug fixes and improvements without requiring manual updates.

Mistral Vibe is available on Le Chat Pro and Team plans with pay-as-you-go credits for extended usage. Users can also bring their own API keys for access.

Pricing structure changes

Devstral 2 has moved to paid API access, priced at $0.40 per million input tokens and $2.00 per million output tokens. Devstral 2 Small costs $0.10 per million input tokens and $0.30 per million output tokens.

Free API usage remains available on the Experiment plan for testing and prototyping. Students receive a 50% discount on Pro plans.

Enterprise offerings

Mistral AI offers enterprise add-ons including fine-tuning on internal languages and domain-specific languages, reinforcement learning with custom environments, and end-to-end code modernization services. The company said it already delivers these solutions for organizations in finance, defense, and infrastructure.

The Vibe CLI can be installed via curl script or uv tool install. Developers need to sign up through Mistral Studio to unlock full access to the coding agent.