Connect OpenAI

1. Open the OpenAI platform and create an API key with access to the intended project. Confirm billing and quotas; a ChatGPT subscription does not replace API access.
2. In Browser Mate, open Settings → LLM Model and select OpenAI.
3. In Model name, enter a model available on your account that supports Chat Completions and tools. gpt-4o-mini is a suggestion present in the runtime, not a guarantee of future availability.
4. The default endpoint is https://api.openai.com/v1/chat/completions, not editable for this provider.
5. Enter the full key. The runtime validates sk-… or sk-proj-… formats; do not use an Anthropic or OpenRouter key.
6. Click Save, then Test connection. The test can be billed.
7. Try “Summarize this page without performing actions”. Only afterwards test automation on a page without consequences.

If it doesn’t work

Check the model identifier, the key’s permissions and the balance. Models exclusive to another API are not automatically compatible with this endpoint. For an OpenAI-compatible gateway, use Custom instead of trying to change the fixed URL.

If you use images, check the capabilities before enabling screenshots. Requests go directly to OpenAI; see the provider’s policy and Privacy.