Free
Limited messages, web searches, coding sessions. Access to Mistral SOTA models.
Open-source agentic coding assistant by Mistral — terminal, IDE, and async agents powered by Devstral 2.
Mistral Vibe is Mistral AI's open-source agentic coding assistant that runs as a CLI, IDE extension, and remote agent. Powered by Devstral 2, Codestral, and Codestral Embed, it edits files, executes shell commands, orchestrates subagents, and extends through skills and MCP servers. Launched December 2025, it now includes async agents, remote agents via Mistral Medium 3.5, native VS Code and JetBrains extensions, Zed integration, voice mode, and tab-to-complete — all under an Apache-2.0 license.
Limited messages, web searches, coding sessions. Access to Mistral SOTA models.
All-day coding in CLI/IDE/web, more messages and web searches, image generation, chat/email support.
Up to 30GB storage/user, domain verification, data export, collaborative workspace.
Custom models, custom agents, SAML SSO, audit logs, white label, private deployments.
Mistral Vibe is Mistral AI's open-source agentic coding assistant — terminal-native, IDE-integrated, and cloud-capable, powered by Devstral 2, Codestral, and Codestral Embed. Launched on December 8, 2025 under an Apache-2.0 license, it started as a CLI-only tool and now runs in your terminal, as a VS Code or JetBrains extension, via the Zed editor, and as a remote agent in Le Chat.
Vibe reads and edits files, executes shell commands, browses codebases, orchestrates subagents for parallel work, and extends its capabilities through a skills system and MCP server configuration.
It's the European sovereign alternative to US-hosted coding agents.
Devstral 2 runs via Mistral API or as a self-hosted open-weight model.
As of May 2026 (v2.10.1, 52 releases), Mistral Vibe supports async agents for work that persists beyond your session, remote agents powered by Mistral Medium 3.5, tab-to-complete for multi-line code completions, voice mode, and AGENTS.md custom instructions. It is available on Pro, Team, API, and Enterprise plans.
/teleport--agent and add project-specific or user-level instructions action.yml in the repository for CI/CD automationThe core of Mistral Vibe is a conversational CLI built for software engineering. It reads, writes, and patches files, executes shell commands in a stateful terminal, recursively searches code with ripgrep, manages a todo list, and asks interactive clarifying questions. The CLI features autocompletion for slash commands and file paths, persistent command history, beautiful themes, and keyboard shortcuts for tool output, todo view, and debug console.
Vibe supports multiple built-in agent profiles: default (requires tool approval), plan (read-only for exploration), accept-edits (auto-approves file changes), and auto-approve (full autonomy). You can create custom profiles with TOML configuration files.
Vibe's subagent system lets the main agent delegate tasks to specialised sub-agents that run independently. The built-in explore sub-agent handles codebase exploration, and you can create custom sub-agents with agent_type = "subagent" in your configuration. Sub-agents prevent context overload by offloading parallel work — one sub-agent can research while another implements, while a third runs tests.
Vibe follows the Agent Skills specification for skill format and structure. Skills are defined in directories with a SKILL.md file containing YAML frontmatter metadata. They can add new tools, custom slash commands, and specialised behaviours. Skills are discovered from multiple locations: custom paths, the project's .vibe/skills/ directory, the standard .agents/skills/ path, and the global ~/.vibe/skills/ directory.
Mistral Vibe ships native extensions for VS Code and JetBrains IDEs with intelligent code completions, context-aware chat, and inline code editing directly in the editor. For Zed, integration is available via the Agent Client Protocol (ACP), with setup instructions in the repository . Vibe is no longer a terminal-only tool — developers can work entirely from their preferred IDE.
Async agents let Vibe keep working after you close the session — useful for long-running tasks like migrations, large-scale refactoring, or batch testing. Remote agents, announced April 29, 2026, are powered by Mistral Medium 3.5 and run in a cloud sandbox within Le Chat via the Vibe Code Workflow. You can start a session from Le Chat, send a prompt from the CLI with & <prompt>, or /teleport an in-progress local session to the cloud.
Mistral Vibe uses a trio of Mistral's coding models:
All three can be accessed via Mistral API or self-hosted, giving enterprises full control over data residency and infrastructure.
Vibe supports the Model Context Protocol (MCP) for connecting to external tools and data sources. Configure MCP servers via config.toml with HTTP, streamable HTTP, or stdio transports. MCP tools are named using the {server_name}_{tool_name} pattern and can be configured with permissions like built-in tools. This ecosystem lets Vibe integrate with databases, APIs, documentation sources, and custom internal tools.
Mistral Vibe is available on four tiers:
The Pro plan ($14.99/mo) is the Le Chat Pro subscription, which includes full Vibe CLI access. The CLI is free to install and use with your own API key. The full source code is available on GitHub under Apache-2.0.
Latest release — improvements to async agents, MCP server configuration, and IDE extension stability
Run 'vibe' in your terminal or open the VS Code extension, then: 'Analyse all API routes in this Express app, identify endpoints missing input validation, add Zod schemas for each, write tests for the validation logic, deploy a staging branch, and summarise what you changed across a PR description.' Vibe reads the codebase, plans using Devstral 2, spawns subagents for parallel research and implementation, applies edits, runs tests, creates the branch, and posts a structured summary. With async agents, the same work can continue after you close the session. AI code review platform for the AI era. Automated code reviews, security scanning, and team analytics across GitHub, GitLab, VS Code, and JetBrains. Used by 300,000+ developers.
AI-powered PR description generator and code review assistant. Automatically writes pull request descriptions, sends stakeholder notifications, creates changelogs, and provides inline code refactoring.
Multi-agent AI coding platform with 12+ agents and 24+ models, featuring Chairman LLM for parallel multi-agent evaluation and end-to-end encrypted inference. Ships across six surfaces: CLI, IDE, Cloud, API, Mobile, and Builder.