Configure the LLM

Last updated: September 12, 2025

Choose an LLM Provider

VeloFill works with OpenAI‑compatible endpoints: OpenAI, third‑party hosted APIs, or your own local server.

Set Endpoint and Key

  1. Open the VeloFill Options page.
  2. Enter the API base URL (e.g., https://api.openai.com/v1). Use HTTPS for remote providers. HTTP may be acceptable for localhost during development; ensure CORS is configured correctly.
  3. Enter your API key securely.
  4. Pick a model (e.g., gpt-4o, a compatible alternative, or your local model).
VeloFill options page showing LLM configuration fields
Configure your API endpoint, key, and model on the options page.

Recommendations

  • Use HTTPS endpoints and rotate keys periodically.
  • Prefer models with strong JSON/structured output for highest accuracy.
  • Monitor provider rate limits and quotas to avoid delays.
  • If running locally, confirm CORS and certificates are configured correctly.

Testing the connection and loading models

  1. Click Test Connection. VeloFill will call your provider’s GET /models endpoint.
  2. On success, the Model dropdown appears and is populated in alphabetical order.
  3. Select your preferred model. The selection is saved with your options.

If the test fails, you’ll see an error message (for example, invalid API key, network error, or CORS failure). See Troubleshooting for fixes.

Provider and URL tips

  • API base URL must point to an OpenAI‑compatible root and typically ends with /v1.
  • Header limitations: VeloFill currently sends Authorization: Bearer <key> and Content-Type: application/json only. Providers that require additional custom headers (e.g., organization or project IDs in headers) are not supported.
  • Local/self‑hosted servers must be reachable from the browser and allow cross‑origin requests (CORS).

Security notes

  • Keys are stored locally by the browser extension storage; they are not transmitted to VeloFill servers (we don’t run any).
  • Always use HTTPS for remote providers to protect form and knowledge base data in transit.

Read more in our Privacy Policy. If you need help choosing a provider or configuring a local server, contact us via the Contact page.

Need help?

If connection tests keep failing, check Troubleshooting or reach out on the Contact page with the error message you’re seeing.