Use JetBrains AI Assistant with open-source models hosted in the EU

JetBrains AI Assistant is the AI plugin for IntelliJ IDEA, PyCharm, WebStorm, GoLand and the other JetBrains IDEs. It provides AI Chat with project context, in-editor code generation, explanations, test and documentation generation, commit messages and integrated agents.

By default it uses models provided through the JetBrains AI service. With bring your own key, it can also use third-party providers, including any endpoint compatible with the OpenAI API. That option is how you connect inference.eu and choose which models work on your code.

In Providers & API keys, select OpenAI-compatible under Third-party AI providers, enter the inference.eu base URL ending in /v1 and your key, and test the connection. The models then appear in a separate section of the AI Chat model selector and in the selector of compatible agents.

Unlike OpenAI's own models, models from OpenAI-compatible endpoints are not assigned to features automatically. Until you assign them under Models Assignment, they are available only in AI Chat. Core features covers in-editor generation, commit messages and the default chat model; Instant helpers covers light tasks such as chat titles and name suggestions. Inline code completion is configured separately.

For Core features, use a strong coding model such as qwen3-coder-480b-a35b-instruct; it handles code generation, refactoring suggestions and explanations well. glm-5.3 and kimi-k3 are good alternatives in AI Chat for longer discussions and design questions.

For Instant helpers, pick a fast model such as gpt-oss-120b, since these calls happen often and should return quickly. AI completion features need a model built for fill-in-the-middle completion; general chat models are not suitable there. Keep JetBrains' default completion or turn it off if you have not chosen a dedicated completion model.

If Test Connection fails, check that the URL ends in /v1 and that your network allows outbound HTTPS to the inference.eu host. A 401 response means the key is wrong. If models appear in AI Chat but features such as commit messages stay unavailable, you have not assigned models under Models Assignment yet.

If the provider option is missing or locked, your organisation may manage AI centrally through JetBrains IDE Services or JetBrains Central; ask an administrator to allow third-party providers. When answers ignore tools or project context, try Qwen3 Coder or GLM-5.3, which handle tool calls more reliably.

When an inference.eu model handles a request, that request is processed in EU data centres and never used for training. We sign a data processing agreement under Article 28 GDPR, share our subprocessor list and bill in euros.

Plan the rest of the setup carefully. If you also activate a JetBrains AI subscription, features your models do not cover fall back to JetBrains AI models. AI completion uses JetBrains models unless you configure another completion provider. For the strictest setups, use BYOK without JetBrains AI, assign inference.eu models to all features you need, and disable the features you have not assigned.