Use A Custom OpenAI-Compatible Endpoint
Use a custom OpenAI-compatible endpoint when your provider or proxy exposes an OpenAI-style API.
Required Fields
Section titled “Required Fields”| Field | What To Enter |
|---|---|
| Base URL | The provider or proxy API base URL. |
| API key | The key required by that endpoint. |
| Model name | The exact model identifier the endpoint expects. |
Setup Steps
Section titled “Setup Steps”- Open onboarding or Settings -> LLM Providers.
- Choose the custom provider option.
- Enter the base URL.
- Enter the API key.
- Enter the model name.
- Validate the configuration.
- Run a short test analysis.
Common Mistakes
Section titled “Common Mistakes”- Missing
/v1when the provider expects it. - Using a model display name instead of the API model identifier.
- Pasting a key for the wrong provider or proxy.
- Forgetting that provider retention, billing, and access rules are controlled by the custom endpoint operator.
Custom provider requests can include prompts, transcript context, screenshot-derived context, and custom questions.
Custom Endpoint Questions
Section titled “Custom Endpoint Questions”What makes an endpoint OpenAI-compatible?
Section titled “What makes an endpoint OpenAI-compatible?”The endpoint should accept OpenAI-style API requests for chat or responses using the base URL, key, and model name you enter in ExtraBrain.
What should I test after saving a custom endpoint?
Section titled “What should I test after saving a custom endpoint?”Run a short analysis with non-sensitive transcript or screenshot context. Confirm the endpoint returns a response and that the model name matches what your provider or proxy expects.