How to add a custom OpenAI compatible LLM backend

[Abdulee]

Hello friends,
How can i add a custom OpenAI compatible LLM backend / API.. For example the endpoint is http://example.com/v1

[sueun-dev]

from current main, the main voice-agent LLM path looks like openai + custom base_url.

The form is generated from /api/v1/user/configurations/defaults; OpenAILLMService has a custom-input model plus base_url defaulting to https://api.openai.com/v1, and create_llm_service() passes user_config.llm.base_url into create_llm_service_from_provider(), which forwards it to Pipecat's OpenAILLMService.

So in Model Configurations -> LLM: choose OpenAI, enter your model manually if it isn't in the dropdown, set base_url to http://example.com/v1, and fill the API key your endpoint expects.

Two caveats I noticed: SaaS mode blocks localhost/private/internal base URLs, while OSS mode allows local endpoints. Also the QA/node-summary path currently only forwards base_url for openrouter, not openai, so if QA analysis doesn't hit your custom endpoint, that looks like a separate code issue.