Continue
Open-source IDE assistant that you point at any model you like - including Mistral Codestral or a local server - so you choose where inference happens
Quick Overview
| Company | Continue (open source project) |
|---|---|
| Category | AI Coding Assistants |
| Headquarters | Distributed, Open Source |
| Founded | 2023 |
| EU Presence | Depends on the model provider you configure |
| Data Location | Whatever provider you point it at - EU, US or local |
| Open Source | Yes (Apache 2.0) |
| GDPR Compliant | Depends on configured provider |
| Self-Hosting | Yes - fully local with Ollama or vLLM |
| Main Features | Configurable model backend, autocomplete, chat, agent mode, custom context providers, VS Code and JetBrains plugins |
| Pricing | Free and open source; you pay only for whatever model provider you choose |
| Best For | Developers who want to keep VS Code but route inference to a European or local model |
| Replaces | GitHub Copilot, Cursor |
Detailed Review
Continue does not have an opinion about which model should write your code. It is the plumbing — completion, chat, agent mode, context handling — with the model provider left as a configuration choice. That single design decision is what makes it relevant to a European stack: you can point it at Mistral's Codestral, at a Hugging Face inference endpoint, or at a model running locally through Ollama.
The project is Apache-licensed and fully open source, so the extension itself introduces no vendor relationship at all.
Choosing Where Inference Happens
A short config file maps roles — autocomplete, chat, embeddings — to providers. A common European setup uses Codestral for completion and Mistral Large for chat, keeping both inside EU infrastructure. A stricter setup points everything at a local Ollama server, in which case nothing leaves the machine.
This is the pragmatic answer for teams that want to stay in VS Code, which JetBrains AI Assistant cannot serve, and cannot accept a fixed US provider, which Cursor and Copilot both are.
Context Control
Continue lets you define custom context providers — a folder, a documentation site, a terminal output, the current diff — and reference them explicitly in a prompt. Compared to assistants that guess at relevant context, this gives more predictable results and, importantly, more predictable data exposure: you can see exactly what is being sent.
Governance Note
Continue is developed by a US-incorporated company alongside its open-source community, so the project itself is not European. What it provides is control: because it is Apache-licensed and self-configured, the data path is yours to define, and no telemetry or code needs to reach the project maintainers at all.
Limitations to Consider
Continue asks more of the user than a turnkey product. You configure providers, manage API keys and tune context settings yourself, and a badly chosen model produces disappointing results that are easy to blame on the extension. Its agent mode is behind Cursor's, and the project itself is US-incorporated even though the data path is under your control.
Who Should Use Continue
VS Code teams that want European or local inference without leaving their editor, and anyone who prefers configuring their own stack to accepting a vendor's default. Paired with Codestral it is the most direct way to get a European AI coding assistant into an existing workflow.
Pros and Cons
Pros
- Point it at any model, including EU and local ones
- Fully open source under Apache 2.0
- Explicit, inspectable context control
- Works in VS Code and JetBrains
- No cost beyond your model provider
Cons
- Requires configuration and key management
- Agent mode behind commercial rivals
- Project itself is US-incorporated
Alternatives to Continue
Other European ai coding assistants worth comparing before you decide:
Mistral AI
France · European frontier models with Codestral for code generation
Hugging Face
France · Open model hub with code models you can run yourself
JetBrains AI Assistant
Czech Republic · AI built into the IDEs your team already uses
Lovable
Sweden · Prompt-to-app builder that ships full-stack projects
See all European ai coding assistants or compare them directly against GitHub Copilot on our alternatives page.
Frequently Asked Questions
Configure it to use a European model provider. Mistral's Codestral for autocomplete and Mistral Large for chat keeps inference in the EU; pointing it at a local Ollama server keeps everything on your own machine.
The extension is free and open source under Apache 2.0. Your only cost is whatever model provider you configure, which is zero if you run models locally.
No. The project is developed by a US-incorporated company with an open-source community. What makes it useful here is that it is Apache-licensed and lets you choose a European or local inference path, so no code has to reach any US service.
Yes, and in JetBrains IDEs as well. VS Code is the more complete of the two integrations.