Configure a provider

Vibern hosts no models of its own. Every request is made by the Claude Code install on your machine, so you can use the official account or point at any protocol-compatible service.

Official Claude account

Install the Claude Code CLI and sign in. Vibern picks up the credentials automatically, no extra setup. Verify with:

claude --version

A version number means you are good.

Using an API key

Go to Settings → Providers → New config and fill in:

Field	Notes
Name	Anything recognizable; it shows in the tab bar
Protocol	anthropic or openai
Base URL	Service endpoint, e.g. https://api.anthropic.com
API key	Your key, stored encrypted on disk
Default model	e.g. claude-opus-4

Keys stay in your local config. Vibern never uploads them and never writes them to logs.

OpenAI-compatible services

Pick the openai protocol and Vibern translates between the two formats: requests are converted on the way out, responses converted back into the shape Claude Code expects.

sequenceDiagram
    participant U as You
    participant V as Vibern
    participant B as Bridge
    participant S as Model service
    U->>V: Ask a question
    V->>B: Anthropic-format request
    B->>S: Converted to OpenAI format
    S-->>B: OpenAI-format response
    B-->>V: Converted back
    V-->>U: Streamed output

Switching providers is a config change, not a code change.

Several configs at once

You can keep multiple configs and pick one per session tab. A common split:

• Something fast and cheap for everyday edits.
• Something stronger for larger refactors.
• An internal gateway for company projects.

Click the model name on the right of the tab bar to switch. It only affects the current tab.

When it will not connect

1. curl the base URL from a terminal to confirm the network path.
2. Check the key for stray spaces or newlines.
3. Open Settings → Logs for the full request error.
4. Behind a proxy, confirm the proxy variables reach Vibern.

Still stuck? See the FAQ.