[GH-ISSUE #24295] issue: Reasoning UI shows "Thought for less than a second" too early while streamed reasoning/tool execution is still running (LiteLLM Responses API + Azure OpenAI)

Originally created by @DediCATeD88 on GitHub (May 1, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/24295 ### Check Existing Issues - [x] I have searched for any existing and/or related issues. - [x] I have searched for any existing and/or related discussions. - [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!). - [x] I am using the latest version of Open WebUI. ### Installation Method Docker ### Open WebUI Version v0.9.2 ### Ollama Version (if applicable) _No response_ ### Operating System Debian 12 (Docker host) ### Browser (if applicable) Chrome ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.** - [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc). - [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps: - Start with the initial platform/version/OS and dependencies used, - Specify exact install/launch/configure commands, - List URLs visited, user input (incl. example values/emails/passwords if needed), - Describe all options and toggles enabled or changed, - Include any files or environmental changes, - Identify the expected and actual result at each stage, - Ensure any reasonably skilled user can follow and hit the same issue. ### Expected Behavior When using reasoning models via LiteLLM Responses API and Azure OpenAI Responses API, the Open WebUI reasoning UI should remain in the "thinking" state until the model has actually finished reasoning / tool execution. ### Actual Behavior Open WebUI shows the reasoning block and starts streaming it, but as soon as the first reasoning/content chunks arrive, the UI already changes to something like: `Thought for less than a second` At that point, the model is still actively working in the background: - continuing reasoning - calling tools - executing actions - streaming more output later So the UI indicates that reasoning is already finished, while the request is actually still running. This is very confusing for the user. ### Steps to Reproduce 1. Deploy Open WebUI `v0.9.2`. 2. Connect Open WebUI to a locally hosted LiteLLM proxy. 3. Configure LiteLLM to use Azure OpenAI via the **Responses API**. 4. Use a reasoning-capable model through LiteLLM. 5. In LiteLLM, enable: - `reasoning_auto_summary: true` 6. In Open WebUI, use reasoning with `reasoning_effort=medium`. 7. Send a prompt that triggers visible reasoning and tool usage. 8. Observe the reasoning UI in Open WebUI: - the reasoning block starts streaming - very early, Open WebUI already shows `Thought for less than a second` - however the model continues reasoning / tool calling / working afterwards until the final answer is complete ### Logs & Screenshots No errors visible ### Additional Information Backend chain: `Open WebUI -> LiteLLM -> Azure OpenAI Responses API` This issue appears reproducibly with Open WebUI `v0.9.2` and earlier versions in my environment. The main UX problem is not that reasoning is shown, but that Open WebUI marks the reasoning phase as finished far too early, while work is still ongoing in the background. This makes the reasoning/timer UI misleading for users.

GiteaMirror commented

2026-05-15 16:17:01 -05:00

Owner

Originally created by @DediCATeD88 on GitHub (May 1, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/24295

Check Existing Issues

I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.

Installation Method

Docker

Open WebUI Version

v0.9.2

Ollama Version (if applicable)

No response

Operating System

Debian 12 (Docker host)

Browser (if applicable)

Chrome

Confirmation

I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.

Expected Behavior

When using reasoning models via LiteLLM Responses API and Azure OpenAI Responses API, the Open WebUI reasoning UI should remain in the "thinking" state until the model has actually finished reasoning / tool execution.

Actual Behavior

Open WebUI shows the reasoning block and starts streaming it, but as soon as the first reasoning/content chunks arrive, the UI already changes to something like:

Thought for less than a second

At that point, the model is still actively working in the background:

continuing reasoning
calling tools
executing actions
streaming more output later

So the UI indicates that reasoning is already finished, while the request is actually still running. This is very confusing for the user.

Steps to Reproduce

Deploy Open WebUI v0.9.2.
Connect Open WebUI to a locally hosted LiteLLM proxy.
Configure LiteLLM to use Azure OpenAI via the Responses API.
Use a reasoning-capable model through LiteLLM.
In LiteLLM, enable:
- reasoning_auto_summary: true
In Open WebUI, use reasoning with reasoning_effort=medium.
Send a prompt that triggers visible reasoning and tool usage.
Observe the reasoning UI in Open WebUI:
- the reasoning block starts streaming
- very early, Open WebUI already shows Thought for less than a second
- however the model continues reasoning / tool calling / working afterwards until the final answer is complete

Logs & Screenshots

No errors visible

Additional Information

Backend chain:

Open WebUI -> LiteLLM -> Azure OpenAI Responses API

This issue appears reproducibly with Open WebUI v0.9.2 and earlier versions in my environment.

The main UX problem is not that reasoning is shown, but that Open WebUI marks the reasoning phase as finished far too early, while work is still ongoing in the background.

This makes the reasoning/timer UI misleading for users.

GiteaMirror added the bug label 2026-05-15 16:17:01 -05:00

[GH-ISSUE #24295] issue: Reasoning UI shows "Thought for less than a second" too early while streamed reasoning/tool execution is still running (LiteLLM Responses API + Azure OpenAI) #90993