[feature request] Use openAI/LiteLLM as Embedding Model

[DediCATeD88]

Is your feature request related to a problem? Please describe.
I have several more powerfull instances than the open WebUi Machine itsef. Is it possible to use openAI/LiteLLM as Embedding Model

Describe the solution you'd like
Give the possibility to use openAI/LiteLLM as Embedding Model

[RomainNeup]

Thanks for the issue. Similar to what we discussed in issue #24, it seems like it could be a great addition. And we may be able to build something robust to support as much implementation as we want.

If OpenWebUI supports connecting directly to their implementation, that would make things really easy. Do you know if OpenWebUI supports it?

[DediCATeD88]

Thanks for the issue. Similar to what we discussed in issue #24, it seems like it could be a great addition. And we may be able to build something robust to support as much implementation as we want.

If OpenWebUI supports connecting directly to their implementation, that would make things really easy. Do you know if OpenWebUI supports it?

"connecting directly" -> that's what i tried to achieve with gemini 2.5 pro and the docus from open WebUI. Every time the answer is there is no documentation for this.

So at the moment i went for -> removing rag with gemini 2.5 pro. At the moment i run without rag. Works fine. But better would be "OpenWebUI supports connecting directly to their implementation" you're right.

[GilGedje]

any news about this? connecting openai(vllm) external embedding model