I have searched for all existing open AND closed issues and discussions for similar requests. I have found none that is comparable to my request.
Verify Feature Scope
I have read through and understood the scope definition for feature requests in the Issues section. I believe my feature request meets the definition and belongs in the Issues section instead of the Discussions.
Problem Description
Hi! Nice project! I'm trying to compare different models ability to call my tools and to select which model generates the tool arguments most accurately. I tried arena models but the models are not calling the tools because the tool definitions (MCP tool) are not sent to the back end LiteLLM request (I am using litellm for my LLM gateway to openai compatible LLM servers).
Image showing the LLM not calling the tools. The models are workspace models with a MCP tool added to each model.
I noticed from the LiteLLM logs that the tool definitions are not sent to the chat completions API hence why the LLMs are not calling the tools.
Example of chat completions API request (using gemini 2.5-pro as the base model) not having the tool definitions sent
{"model":"gemini-2.5-pro","stream":true,"tools":[#thisismissingfromthechatcompletionsrequest{type:"function",function:{name:"query_database",parameters:{},description:"..."}}],#thetoolsshouldbeincludedinthechatcompletionsAPIrequest"messages":[{"role":"system","content":"You are a detailed and helpful security AI assistant that helps draws on tools to create reports, conduct data analysis and troubleshooting. You have tools to retrieve more information about the robot mission, alerts event logs from the aria query_database tool. Do not create table if the data is a lot and exceeds more than 10 rows of data, unless the user explicitly instructs to create table to list all the data. The current date time is 2026-02-09 18:03:12. The user's timezone is Asia/Singapore"},{"role":"user","content":"what were the alerts for colab 22 from 1 jan 2026 to 8 jan 2026? Where were these alerts and how many were there at where?"}]}
Desired Solution you'd like
Attach the tool definitions as per normal how the tools work with models in a single chat, but extended for models in arena models mode. This will allow evaluation of tool calling on the different models within a single interface, instead of current approach where I have to open a new chat for each workspace model and send the same prompt to the models individually.
Alternatives Considered
No response
Additional Context
No response
Originally created by @zhiweit on GitHub (Feb 9, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/21271
### Check Existing Issues
- [x] I have searched for all existing **open AND closed** issues and discussions for similar requests. I have found none that is comparable to my request.
### Verify Feature Scope
- [x] I have read through and understood the scope definition for feature requests in the Issues section. I believe my feature request meets the definition and belongs in the Issues section instead of the Discussions.
### Problem Description
Hi! Nice project! I'm trying to compare different models ability to call my tools and to select which model generates the tool arguments most accurately. I tried arena models but the models are not calling the tools because the tool definitions (MCP tool) are not sent to the back end LiteLLM request (I am using litellm for my LLM gateway to openai compatible LLM servers).
Image showing the LLM not calling the tools. The models are workspace models with a MCP tool added to each model.
<img width="1578" height="1117" alt="Image" src="https://github.com/user-attachments/assets/5b22a0f2-6bfb-46e9-a228-e8e297e933b5" />
I noticed from the LiteLLM logs that the tool definitions are not sent to the chat completions API hence why the LLMs are not calling the tools.
Example of chat completions API request (using gemini 2.5-pro as the base model) not having the tool definitions sent
```json
{
"model": "gemini-2.5-pro",
"stream": true,
"tools":[ # this is missing from the chat completions request
{
type:"function",
function:{
name:"query_database",
parameters:{},
description:"..."
}
}
], # the tools should be included in the chat completions API request
"messages": [
{
"role": "system",
"content": "You are a detailed and helpful security AI assistant that helps draws on tools to create reports, conduct data analysis and troubleshooting. You have tools to retrieve more information about the robot mission, alerts event logs from the aria query_database tool. Do not create table if the data is a lot and exceeds more than 10 rows of data, unless the user explicitly instructs to create table to list all the data. The current date time is 2026-02-09 18:03:12. The user's timezone is Asia/Singapore"
},
{
"role": "user",
"content": "what were the alerts for colab 22 from 1 jan 2026 to 8 jan 2026? Where were these alerts and how many were there at where?"
}
]
}
```
### Desired Solution you'd like
Attach the tool definitions as per normal how the tools work with models in a single chat, but extended for models in arena models mode. This will allow evaluation of tool calling on the different models within a single interface, instead of current approach where I have to open a new chat for each workspace model and send the same prompt to the models individually.
### Alternatives Considered
_No response_
### Additional Context
_No response_
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @zhiweit on GitHub (Feb 9, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/21271
Check Existing Issues
Verify Feature Scope
Problem Description
Hi! Nice project! I'm trying to compare different models ability to call my tools and to select which model generates the tool arguments most accurately. I tried arena models but the models are not calling the tools because the tool definitions (MCP tool) are not sent to the back end LiteLLM request (I am using litellm for my LLM gateway to openai compatible LLM servers).
Image showing the LLM not calling the tools. The models are workspace models with a MCP tool added to each model.
I noticed from the LiteLLM logs that the tool definitions are not sent to the chat completions API hence why the LLMs are not calling the tools.
Example of chat completions API request (using gemini 2.5-pro as the base model) not having the tool definitions sent
Desired Solution you'd like
Attach the tool definitions as per normal how the tools work with models in a single chat, but extended for models in arena models mode. This will allow evaluation of tool calling on the different models within a single interface, instead of current approach where I have to open a new chat for each workspace model and send the same prompt to the models individually.
Alternatives Considered
No response
Additional Context
No response