Replies: 13 comments 29 replies
-
|
I have that question too. Great. But actually the generate doesn't work I get According to the documentation it should behave like the original endpoint. If I do reverse-proxy as mentioned here: The Whitelist filter of openwebui is basically useless. |
Beta Was this translation helpful? Give feedback.
-
|
@dbenoot33 how can I add a RAG file that is already processed and part of Open Web UI to the request? I can't find the documentation of the API. Also is this safe for a production environment? Around 10k requests per day. |
Beta Was this translation helpful? Give feedback.
-
|
Is there a way to log the messages/prompts requested via user API in Open Web UI Database? I see in the chat history only the chat logs of users made in GUI and not via the API. I am wondering if it is already implemented? |
Beta Was this translation helpful? Give feedback.
-
|
Thank you! Saved me a lot of time! |
Beta Was this translation helpful? Give feedback.
-
I am facing same problem, i think the way to do it is to set collection name when you first process the document via API. |
Beta Was this translation helpful? Give feedback.
-
This seems to work, but only for models from the offical website and not the models created via Modefile in the WebUi. |
Beta Was this translation helpful? Give feedback.
-
|
You can use Ollama's API Server to make request as this is the llama.cpp wrapper used. |
Beta Was this translation helpful? Give feedback.
-
|
No doubt, but how to use the OpenwebUI API Server with the individual API key, including the usage of the whitelist of OpenwebUI? |
Beta Was this translation helpful? Give feedback.
-
|
... and maybe it is possible to use the Open AI models as well (if the OpenAi key is entered in OpenWeb UI? |
Beta Was this translation helpful? Give feedback.
-
|
Is there way to create WebUI API keys automatically? I am trying to get the user message and details like files selected, collection names from a question by a user for my custom RAG Agent pipeline and currently passing the API keys manually, is there a way I can get it created automatically for every user? |
Beta Was this translation helpful? Give feedback.
-
|
Routing is really weird for OpenAI endpoints I would like to use OpenWebUI as an LLM router, API keys creation interface is really convenient, it's nice for users to have a single point of entry, but the routing needs to be fixed. |
Beta Was this translation helpful? Give feedback.
-
|
I don't mind the dual-stack, I mean, you can avoid it right now by registering ollama as an openai endpoint. I just wish we could route properly between different openai endpoints and their models. Right now the best workaround is to have an additional LiteLLM layer I guess. |
Beta Was this translation helpful? Give feedback.
-
|
We already went down the integrated LiteLLM road and things didn't work out, best we can do there is rolling it into Pipelines or hitting it directly as its own standalone container. |
Beta Was this translation helpful? Give feedback.
-
|
nvm, just went through https://github.com/open-webui/open-webui/blob/main/backend/main.py#L1106-L1112 |
Beta Was this translation helpful? Give feedback.
-
|
Beta Was this translation helpful? Give feedback.
-
|
@justinh-rahb So this works well for LLM endpoints, but if I wanted to proxy something like https://github.com/fedirz/faster-whisper-server through Open WebUI, I'm out of luck. I'd like owui to be a point of entry for users so they can use settings page to generate an API key and use it for Right now there's no way to have a single API key for all of those (and /audio is only usable internally as far as I see, there's no way to be compatible with OPEANAI_BASE_URL in openai's bindings). One solution right now is to use LiteLLM and manage keys there, but they're greedy with putting SSO behind enterprise paywall. I'd love for Open WebUI to be my main model router for all use-cases. |
Beta Was this translation helpful? Give feedback.
-
|
I am not sure if I hit a bug or if it's a PBKAC... This works: This does not work: The full error I see in the logs says: |
Beta Was this translation helpful? Give feedback.
-
|
I have tried stream false and JSON mode with chat endpoint and had similar if not the same issue. |
Beta Was this translation helpful? Give feedback.
-
|
I'm currently facing the same issue where I want files automatically send to openwebui from an external website but I can't find anything about rag api endpoints |
Beta Was this translation helpful? Give feedback.
-
|
@kw1c-TimoCorstjens You should check out the OpenAPI Endpoints https:///rag/api/v1/docs and https:///api/v1/docs. The latter allows you to upload files and create docs, and the rag api allow you - among other things - to process previously uploaded files. You can also create an API Token in the UI with "Settings -> Account" and use it as Auth Header. Examples: Hope this helps, I also had to play around a little until the document was finally visible in the UI. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your reply. I will definitely try playing around with this for a while. This is an issue I've been stuck on for a while so I greatly appreciate it! |
Beta Was this translation helpful? Give feedback.
-
|
I am using a browser add-in Page Assist that wants to connect to an Ollama back end. I have set up OpenWebUI as a router between multiple instances of Ollama and I would prefer to connect the browser extension to OpenWebUI if possible. From what I've read OpenWebUI should expose an API endpoint that is functionally equivalent to Ollama. I've tried every combination of URL that I can think of and so far nothing has worked. Any ideas? |
Beta Was this translation helpful? Give feedback.
-
|
You can view all Ollama API endpoints at http://your-ip:3000/ollama/docs |
Beta Was this translation helpful? Give feedback.
-
|
I've found a lot of ways to get 404 and 405 errors ;-) |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
Which endpoint are you hitting? |
Beta Was this translation helpful? Give feedback.
-
|
I'm currently using rag/api/v1/query/doc |
Beta Was this translation helpful? Give feedback.
-
|
Is it possible to chat with our workspace models seen through the Example model I want to chat with from other apps: {
"id": "mister",
"name": "Mister",
"object": "model",
"created": 1725028456,
"owned_by": "ollama",
"info": {
"id": "mister",
"user_id": "ee382672-f37c-433f-8bcb-02c0ce148b73",
"base_model_id": "mistral-nemo:latest",
"name": "Mister",
"params": {},
"meta": {
"profile_image_url": "/static/favicon.png",
"description": "",
"suggestion_prompts": []
},
"updated_at": 1725028456,
"created_at": 1725028456
},
"preset": true
} |
Beta Was this translation helpful? Give feedback.
-
|
I am looking for this as well. I created a custom model through workspaces for a chatbot I want to embed into an app I am building. I cannot figure out how to interact with that model. Any help would be much appreciated. |
Beta Was this translation helpful? Give feedback.
-
|
hello everyone, this topic is interesting me as well because I also try to connect to different models (external models, not with ollama) within the openwebui API. Models are accessible from the interface, now I need to connect using the GET/POST API command. I tried to understant how to combines commands from here : But it is completely unsuccessfull. Is there a basic examples of GET / POST commands where it is explain how to : thank you for your time sharing I'm using the request library from python is it a problem ? |
Beta Was this translation helpful? Give feedback.
-
|
How to query openwebui (assuming Ollama backend) Copy this to a notebook to modify it: curl -X POST -H "Authorization: Bearer ” -H 'Content-Type: application/json' http://localhost:3000/ollama/api/generate -d '{ "model": “”, "prompt": "Why is the sky blue?", "stream": false }' replace localhost:3000 with the correct url (the same host:port you use to browse) Go to settings, account. Copy your jwt token: |
Beta Was this translation helpful? Give feedback.
-
|
can you break this down a bit more please |
Beta Was this translation helpful? Give feedback.
-
please break this down more |
Beta Was this translation helpful? Give feedback.
-
|
how does ollama append that to a current chat? |
Beta Was this translation helpful? Give feedback.
-
|
I am also looking for this feature. For me, OpenWebUI is the public gate of my Ollama in my HomeLab. I do not like to expose the Ollama's port to the WAN or put it back of the Nginx. I am using some app which integrated the "GPT" features if I provide the my API. Now I can only using them when I back to home in the LAN. |
Beta Was this translation helpful? Give feedback.
-
|
I was wondering the same as you, guys: I was looking for an api doc everywhere but, finally, the simplest is only to open your browser dev tool console. Hope it helps others For exemple, steps to create a chat, add a context file on it, and chatI assume you have Open Webui - > localhost:3000, all queries must have the authorization header
And finally (a little bit boring to build the payload for having files but u have all the information with the previous posts)
|
Beta Was this translation helpful? Give feedback.
-
|
After taking a look, open-webui guys are doing an amazing job! File chunks are managed for us, history is simple to maintain, call to the web search method is simple as well and so on (haven't seen for now if at some point they switch to a summary of the discussion to save tokens) But wow! Deserve definitely to have an api wrapper to highlight the backend job or even better, to split the backend into a functional package lib and a simple http wrapper to it |
Beta Was this translation helpful? Give feedback.
-
|
how do you link an ollama response to an existing chat? I have tried many things but it doesnt do it. thanks |
Beta Was this translation helpful? Give feedback.
-
|
there really should be a POST /api/chats/generate that creates a new chat, uses the model specified, with the prompt and documents specified. |
Beta Was this translation helpful? Give feedback.
-
|
I had a similar problem with authentication to my WebUI server and referencing uploaded documents yesterday, finally got it to work. Here is the code that worked for me: |
Beta Was this translation helpful? Give feedback.
and others
-
Is there a way to use OpenWebUI as an API endpoint?
For example do similar request via API to the ones we do on the UI?
Including on the call references to uploaded documents. I can't find on the documentation anything for the API.
Beta Was this translation helpful? Give feedback.
All reactions