Skip to main content
Mistral Vibe+18 Mixed

GILD Review

GILD's call and the buying facts behind it.

From GILD research

Should we adopt Mistral Vibe?

Is Mistral Vibe good enough to replace Claude Code or Codex for your team?

  1. 30-Second TakePick Mistral Vibe as a cheap, supervised terminal agent from a European vendor, and a poor substitute for Claude Code on long tasks.
  2. Best fitTeams that want an Apache 2.0 terminal agent, model traffic with a European provider, and a flat monthly price for supervised, scoped changes.GILD Review
  3. Consider another workflowFor long multi-step changes across a large codebase, Claude Code or Codex fits better; for any model behind one open-source agent, OpenCode or Aider fits better.GILD Review

Drawn from 85 sources

GILD analysis

Mistral Vibe Review

Mistral Vibe gives a team an open-source terminal agent and Mistral-hosted models for US$14.99 a month on Pro, with local models as a fallback. It suits scoped work an engineer watches and teams that want a European vendor, and it trails Claude Code and Codex when tasks run long or the request is vague. Settle plan usage and the training opt-out before more than a few engineers depend on it.

  • AI Coding Agents
  • Engineering Productivity

The decision

Best fit
Teams that want an Apache 2.0 terminal agent, model traffic with a European provider, and a flat monthly price for supervised, scoped changes.
Consider another workflow
For long multi-step changes across a large codebase, Claude Code or Codex fits better; for any model behind one open-source agent, OpenCode or Aider fits better.
Adoption condition
Opt out of model training, set enable_telemetry to false, keep accept-edits or a stricter mode, and turn on pay-as-you-go only with a spending limit.

Basis: Mistral pricing and Vibe documentation checked Oct 25, 2026, plus the public accounts in Signals. We ran no performance benchmark.

Buying & deployment factsWhat engineering leaders should verify before a broader rollout.
How you buy it

US$14.99

A month for Pro before tax. Team is US$24.99 per user, and the Apache 2.0 CLI is free.

Source 5, GitHub, Mistral Vibe license (Apache 2.0): Mistral Vibe license (Apache 2.0)Source 1, Mistral AI, Mistral pricing: Mistral pricingSource 2, Mistral AI, API keys and profiles: API keys and profiles

Details

The CLI is free under Apache 2.0. Hosted use comes with a Mistral plan: Free has limited coding sessions, Pro is US$14.99 per month and Team US$24.99 per user per month before tax, and Enterprise is priced through sales. A Mistral API key or another provider's key also works.

Cost drivers

One shared budget

Vibe Code, Studio, and the API draw on it. Past it, pay-as-you-go bills at API rates.

Source 2, Mistral AI, API keys and profiles: API keys and profiles

Details

A plan's included usage is shared across Vibe Code, Studio, and the API. When it runs out, Vibe Code stops until the next billing period unless pay-as-you-go is on, which draws from prepaid credits.

Where it runs

Terminal, IDE, or local

A CLI for macOS and Linux, VS Code, JetBrains and Zed through ACP, or your own model server.

Source 6, docs.mistral.ai, Install and setup: Install and setupSource 7, docs.mistral.ai, Use Vibe in other IDEs: Use Vibe in other IDEsSource 8, docs.mistral.ai, Vibe Code Web: Vibe Code WebSource 3, Mistral AI, Using offline models: Using offline models

Details

A Python 3.12 CLI whose primary platforms are macOS and Linux (it also runs on Windows), a VS Code extension, JetBrains IDEs and Zed through ACP, and Vibe Code Web sandboxes. Local models run behind vLLM, llama.cpp, LM Studio, Ollama, or any OpenAI-compatible server.

What to verify

Training on by default

Opt out of training and telemetry. The default agent edits files without asking.

Source 4, Mistral AI, Safety, approvals, and permissions: Safety, approvals, and permissionsSource 9, Mistral AI, Configuration: ConfigurationSource 10, Mistral AI, Training on user data (Mistral help): Training on user data (Mistral help)Source 11, Mistral AI, Regional inference: Regional inference

Details

The default agent applies file edits without asking and asks before shell commands. Telemetry is on by default, and input and output on Vibe plans other than Enterprise train Mistral's models unless you opt out. Mistral documents the EU regional endpoint for the API only, and says nothing of it for Vibe.

Sourced from Mistral documentation