Mistral Code

What Is Mistral Code?

Every European developer using GitHub Copilot faces the same uncomfortable fact: their code snippets travel to Microsoft's Azure infrastructure, processed under US cloud terms, stored in data centres that may fall outside EU jurisdiction. For individual developers building side projects, the risk is theoretical. At regulated EU companies — financial services, healthcare, legal tech — it creates real compliance friction with procurement, legal, and data protection officers.

Mistral Code is the answer that Paris-based Mistral AI built. Launched in 2025, it is now the coding offering within Mistral Vibe, the agent formerly called Le Chat. It runs as a terminal-native CLI and as extensions for VS Code, JetBrains IDEs and Zed, with a structural advantage that US competitors cannot match. By default, inference runs on Mistral's EU-hosted infrastructure, under French law, and Enterprise workspaces are opted out of model training by default.

Mistral's product page lists three model families behind it. Mistral Medium, which Mistral calls its most performant model family for complex, multi-step software engineering, handles agentic work alongside Devstral 2. Codestral handles fast inline completions, and Codestral Embed powers semantic code search. Note that Mistral's model documentation lists Devstral 2 as retired from the API on 31 July 2026, with Mistral Medium 3.5 as its replacement.

Mistral AI itself is Europe's most prominent AI lab. In September 2026 it closed a €3 billion Series D led by Samsung Electronics, with the Scaleup Europe Fund and PSG Equity as co-leads, at a post-money valuation above €21 billion. Mistral Code is not a side project — it's a product line from a company with genuine model research depth.

Key Features

Codestral: Inline Completion

Codestral is Mistral's code-specialised model, trained specifically for fill-in-the-middle completion tasks. On the HumanEval and MultiPL-E benchmarks, it scores competitively against GitHub Copilot's backend models for Python, TypeScript, Go, and Rust. In practice, this means fast single-keystroke completions that require minimal manual correction on common patterns.

The inline suggestion experience in both VS Code and JetBrains is close to Copilot's — grey ghost text appearing as you type, accepted with Tab. Latency on the EU servers is low enough that completions arrive within 200-400ms for most file sizes, comparable to Copilot for users in Western Europe where round-trip times to EU servers are short.

Where Codestral shows visible improvement over Copilot is in European-language codebases that use non-English comments and variable names — French, German, Spanish. Mistral's multilingual training means the model understands context that English-optimised models sometimes misread.

Agentic Coding

Agentic mode is where you describe multi-step tasks in natural language. With read and write access to your filesystem, a shell, and a configurable set of tools, it can read files, run commands, write code, and open pull requests on your behalf, under your supervision. Proposed changes arrive as diffs for review rather than silent edits.

A concrete example: given a legacy REST controller in Java and an instruction to refactor it to Spring WebFlux reactive patterns, the agent reads the file and identifies all blocking calls. It then generates replacements across the controller and its test class, and presents both as a unified diff. This saves 20-40 minutes of mechanical work that would otherwise fall to the developer.

On large tasks touching 10 or more files, the agent can feel slower than the leading US agentic editors. For typical three-to-five file refactors, the gap is small enough that EU compliance requirements justify it.

Terminal, IDE and Web

Mistral Code installs as a terminal CLI (via a one-line installer or uv tool install mistral-vibe), as a VS Code extension, and in JetBrains IDEs and Zed. JetBrains support works through the Agent Client Protocol: you install the AI Assistant plugin (version 2025.3.2 or later) and add Vibe from the ACP agent registry. Vibe Code Web runs sessions against a GitHub repository in a remote sandbox. The CLI also gives Vim, Neovim and Emacs users a way in, although there is no first-party extension for those editors.

Setup involves signing in with a Mistral account, where usage and rate limits follow your plan and pay-as-you-go settings, or entering a Mistral Studio (formerly La Plateforme) API key. Requests go to Mistral's EU-hosted API by default. Mistral also offers a US API endpoint, which is used only if you explicitly select it; regional EU and US endpoints became generally available in August 2026.

Mistral Studio Integration

Because Mistral Code accepts standard Mistral Studio API keys, developers and organisations already using Mistral's API for other purposes (LLM inference, embeddings, agents) can consolidate billing and usage monitoring. Token consumption across Mistral Code and any other Mistral Studio usage appears in a single dashboard. For finance teams, this is a meaningful operational simplification.

Pricing

Mistral Code is not sold as a standalone subscription. It is included in Mistral Vibe plans, and it can also run on pay-as-you-go Mistral Studio API tokens.

Vibe Free includes limited coding sessions, enough to test the CLI or an IDE extension. Pro costs €14.99 a month and covers all-day coding in the CLI, IDE or on the web, with pay-as-you-go when you need more and $15 of API credits included each month. Team costs €24.99 per user per month, with a $50 monthly minimum and higher limits for small development teams.

Developers who prefer metered billing can use a Mistral Studio API key instead. Codestral is listed at $0.30 per million input tokens and $0.90 per million output tokens. Agentic sessions cost more per task than completions because reasoning generates longer chains of tokens, so teams running agents at scale should watch consumption in the Mistral Studio dashboard.

Enterprise pricing is custom. It covers custom deployments, including self-hosting Mistral Vibe or deploying on Mistral Compute or major cloud providers, plus fine-tuning on your own codebase.

EU Compliance and Privacy

Mistral Code's compliance profile is straightforward for EU organisations because Mistral AI SAS is French by incorporation. There is no US parent company and no CLOUD Act exposure for data processed through their API, and data stays in the EU by default unless you choose the US endpoint.

Inference for Mistral's coding models runs in EU data centres by default. Training is a setting, not a blanket ban. Mistral's Commercial Terms (effective 25 September 2026) rule out training on customer data or outputs unless training is switched on for that product, you give feedback, or you use Labs or Preview models. Its help centre says free-mode API data may be used for training unless you opt out, pay-as-you-go customers can opt out at any time, and Enterprise is opted out by default. Checking that setting before rollout addresses the concern that killed GitHub Copilot adoption at many European enterprises — the fear that proprietary code becomes training data for models serving competitors.

Mistral AI is actively working on EU AI Act compliance, as models such as Mistral Medium and Codestral qualify as general-purpose AI models under the Act's definitions. French regulators are familiar with the company, reducing the jurisdictional uncertainty that applies to US-based AI providers.

ISO 27001 certification is not yet listed for the Mistral Code product specifically, but Mistral's general infrastructure certifications cover the same servers. Organisations requiring specific certifications should request the latest compliance documentation from Mistral's enterprise team.

Who It's Best For

EU regulated enterprises where DPOs have blocked or restricted GitHub Copilot or Cursor due to US data transfer concerns. Mistral Code's EU hosting by default and Enterprise's default training opt-out address the two most common objections raised in compliance reviews.

Teams already using Mistral's API for LLM or agent workloads. Adding Mistral Code consolidates AI tooling under one vendor, one DPA, one billing account, and one set of procurement docs.

Python, TypeScript, and Rust developers where Codestral's benchmark performance is strongest. Developers primarily working in legacy languages like COBOL or Fortran will see less benefit.

Individual developers who want a terminal agent and IDE extension from one EU vendor, starting on Vibe Free's limited coding sessions and moving to Pro at €14.99 a month for all-day use.

The Verdict

Mistral Code is not yet the complete package that GitHub Copilot represents after five years of development. The lack of first-party Vim, Neovim and Emacs extensions and the slower agentic mode on very large tasks are real gaps. But for developers and organisations who prioritise EU data sovereignty, Mistral Code is the only EU-native coding assistant with frontier-class underlying models. If your work involves code that cannot leave EU infrastructure, Mistral Code goes from "one option among many" to "the obvious choice."

Frequently Asked Questions

Is Mistral Code GDPR compliant?

Yes. By default, inference runs on Mistral's EU-hosted infrastructure, and data is hosted in the United States only if you explicitly select Mistral's US API endpoint. Mistral AI SAS is incorporated in France and subject to GDPR by default. Training on your code is controlled by settings rather than ruled out. Free-mode API data may be used for training unless you opt out, pay-as-you-go customers can opt out at any time, and Enterprise is opted out by default. Enterprise customers can also arrange dedicated VPC or on-premise deployments for maximum isolation.

Which IDEs does Mistral Code support?

Mistral Code runs in the terminal as the Vibe CLI and supports VS Code, JetBrains IDEs and Zed. In JetBrains IDEs such as IntelliJ IDEA, PyCharm, WebStorm and GoLand, it works through the Agent Client Protocol in the AI Assistant plugin. A web version, Vibe Code Web, runs sessions on GitHub repositories in a remote sandbox. There is no first-party Vim, Neovim or Emacs extension, so users of those editors work through the CLI.

How much does Mistral Code cost?

Mistral Code is included in Mistral Vibe plans rather than sold separately. Vibe Free includes limited coding sessions, Vibe Pro costs €14.99 a month for all-day coding, and Vibe Team costs €24.99 per user per month with a $50 monthly minimum. Developers can instead use a Mistral Studio API key and pay per token, with Codestral listed at $0.30 per million input tokens and $0.90 per million output tokens. Enterprise pricing is custom.

How does Mistral Code compare to GitHub Copilot?

Codestral is competitive on fill-in-the-middle benchmarks, and Mistral Code's EU compliance story is significantly stronger — a French company, EU hosting by default, and Enterprise workspaces opted out of model training by default. GitHub Copilot has a larger integration ecosystem, more mature enterprise rollout tooling, and broader IDE support. Copilot is better for teams where compliance is not the priority; Mistral Code is better for EU-regulated organisations.

Where is my code processed when using Mistral Code?

By default, code snippets are processed on Mistral's EU-hosted infrastructure; data goes to the United States only if you explicitly select Mistral's US API endpoint. Whether code can be used for training depends on your plan and settings, so check the training opt-out on your organisation before rollout; Enterprise is opted out by default. Enterprise deployments can use dedicated VPC or on-premise Mistral inference for complete infrastructure isolation.