AI Configuration - auth header support?

[goolsbot]

My environment has api support for a few LLMs we can tap into, but our api requires that we send out our API token in a http header.

For example in curl, we would use something like:

curl -X POST https://url/v1/chat/completions -H "Authorization: Bearer " .... rest of curl information.

I haven't been able to get it to work by adding user/pass, or 'https://tokenid:token@url/v1/chat/....

[dev-hari-prasad]

pgAdmin already supports OpenAI-compatible endpoints. If the said environment is OpenAI-ready, you can simply swap the base URL in AI Preferences and call it a day.

However, if the said environment uses a different request/response format, it would be a significant effort to add support for a non-standard system in pgAdmin.

OpenAI-compatible endpoints are already an industry standard, and virtually all major LLM providers like Gemini, DeepSeek, Anthropic, Ollama, LM Studio, and others support them, it's everywhere.

You should instead check for the compatibility and try to get your environment/provider/endpoint to support OpenAI compatibility.