[GH-ISSUE #18397] issue: 0.6.33 problem with long contest 128k #18584

New Issue

GiteaMirror · 2026-04-20T00:48:07-05:00

GiteaMirror commented

2026-04-20 00:48:07 -05:00

Originally created by @batot1 on GitHub (Oct 17, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/18397

Check Existing Issues

I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I am using the latest version of Open WebUI.

Installation Method

Docker

Open WebUI Version

0.6.33

Ollama Version (if applicable)

0.12.5

Operating System

Debian 12 (all updates)

Browser (if applicable)

FF 143.0.4 (64-bit)

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

BUG long contest 128k.
Past ~second question he never stooped answer and repeats the same thing over and over in an infinite loop.

How reproduce:
/ollama pull qwen3-coder:30b
/set parameter num_ctx 131072
/save qwen3-coder-128k

Now when I'm asking him 2-3 query in window, he getting loop and never stoping answer.
When I'm doing this same direct in window ollama all working property.

Actual Behavior

BUG long contest 128k.
Past ~second question he never stooped answer and repeats the same thing over and over in an infinite loop.

How reproduce:
/ollama pull qwen3-coder:30b
/set parameter num_ctx 131072
/save qwen3-coder-128k

Now when I'm asking him 2-3 query in window, he getting loop and never stoping answer.
When I'm doing this same direct in window ollama all working property.

Steps to Reproduce

How reproduce:
/ollama pull qwen3-coder:30b
/set parameter num_ctx 131072
/save qwen3-coder-128k

Now when I'm asking him 2-3 query in window in OpenWeb-UI, he getting loop and never stoping answer.

Logs & Screenshots

In windows Open-Web-UI Contest:
Response payload is not completed: <TransferEncodingError: 400, message='Not enough data to satisfy transfer length header.'>

Additional Information

In LOG docker I don't see any warring or error only INFO. Nothing special.

Originally created by @batot1 on GitHub (Oct 17, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/18397 ### 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 am using the latest version of Open WebUI. ### Installation Method Docker ### Open WebUI Version 0.6.33 ### Ollama Version (if applicable) 0.12.5 ### Operating System Debian 12 (all updates) ### Browser (if applicable) FF 143.0.4 (64-bit) ### 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 BUG long contest 128k. Past ~second question he never stooped answer and repeats the same thing over and over in an infinite loop. How reproduce: /ollama pull qwen3-coder:30b /set parameter num_ctx 131072 /save qwen3-coder-128k Now when I'm asking him 2-3 query in window, he getting loop and never stoping answer. When I'm doing this same direct in window ollama all working property. ### Actual Behavior BUG long contest 128k. Past ~second question he never stooped answer and repeats the same thing over and over in an infinite loop. How reproduce: /ollama pull qwen3-coder:30b /set parameter num_ctx 131072 /save qwen3-coder-128k Now when I'm asking him 2-3 query in window, he getting loop and never stoping answer. When I'm doing this same direct in window ollama all working property. ### Steps to Reproduce How reproduce: /ollama pull qwen3-coder:30b /set parameter num_ctx 131072 /save qwen3-coder-128k Now when I'm asking him 2-3 query in window in OpenWeb-UI, he getting loop and never stoping answer. ### Logs & Screenshots In windows Open-Web-UI Contest: Response payload is not completed: <TransferEncodingError: 400, message='Not enough data to satisfy transfer length header.'> ### Additional Information In LOG docker I don't see any warring or error only INFO. Nothing special.

GiteaMirror added the bug label 2026-04-20 00:48:07 -05:00

GiteaMirror closed this issue

2026-04-20 00:48:08 -05:00

GiteaMirror commented

2026-04-20 00:48:10 -05:00

@Classic298 commented on GitHub (Oct 17, 2025):

Do you have the hardware to support a 128k context* window?

Reproduction steps not clear; how to reproduce and what to do

@Classic298 commented on GitHub (Oct 17, 2025): Do you have the hardware to support a 128k context* window? Reproduction steps not clear; how to reproduce and what to do

GiteaMirror commented

2026-04-20 00:48:11 -05:00

@batot1 commented on GitHub (Oct 17, 2025):

I was giving you instruction how reprocude:

/ollama pull qwen3-coder:30b
/set parameter num_ctx 131072
/save qwen3-coder-128k

When you saveing this you get "new model with long contest. But it only working wiht ollama but with OpenWebUI not working property.

Yes, I have hardware to reproduce 128k contest.

$ ollama ps
NAME ID SIZE PROCESSOR CONTEXT UNTIL
qwen3-coder-30b-128k:latest 571f59fefc54 44 GB 48%/52% CPU/GPU 131072 4 minutes from now

I'm guessing that OpenWebUI probably won't work with any longer models, as I see they're probably not handling long contexts correctly.
Why are you closing the ticket without resolving or even verifying that the problem exists?
4 simple steps to reproduce.

@batot1 commented on GitHub (Oct 17, 2025): I was giving you instruction how reprocude: /ollama pull qwen3-coder:30b /set parameter num_ctx 131072 /save qwen3-coder-128k When you saveing this you get "new model with long contest. But it only working wiht ollama but with OpenWebUI not working property. Yes, I have hardware to reproduce 128k contest. $ ollama ps NAME ID SIZE PROCESSOR CONTEXT UNTIL qwen3-coder-30b-128k:latest 571f59fefc54 44 GB 48%/52% CPU/GPU 131072 4 minutes from now I'm guessing that OpenWebUI probably won't work with any longer models, as I see they're probably not handling long contexts correctly. Why are you closing the ticket without resolving or even verifying that the problem exists? 4 simple steps to reproduce.

GiteaMirror commented

2026-04-20 00:48:11 -05:00

@Classic298 commented on GitHub (Oct 17, 2025):

You did not provide any sensible steps to reproduce? How do I reproduce these steps inside of Open WebUI?

And yes, Open WebUI CAN handle long context models just fine.

@Classic298 commented on GitHub (Oct 17, 2025): You did not provide any sensible steps to reproduce? How do I reproduce these steps inside of Open WebUI? And yes, Open WebUI CAN handle long context models just fine.

GiteaMirror commented

2026-04-20 00:48:13 -05:00

@batot1 commented on GitHub (Oct 17, 2025):

Open-WebUI side:
new chat ---> search model (select qwen3-coder-30b-128k:latest)

No matter abut you asking example:
"10 rows with question"
--->>> PUSH ENTER
waitting for answer
"10 rows with question"
--->>> PUSH ENTER
waitting for answer
"10 rows with question no matter what you wrote"
--->>> PUSH ENTER
waitting for answer. Probably newer ending but if enndig repeat one again else.
"10 rows with question no matter what you wrote"
--->>> PUSH ENTER

@batot1 commented on GitHub (Oct 17, 2025): Open-WebUI side: new chat ---> search model (select qwen3-coder-30b-128k:latest) 1. No matter abut you asking example: "10 rows with question" --->>> PUSH ENTER 2. waitting for answer 3. "10 rows with question" --->>> PUSH ENTER 4. waitting for answer "10 rows with question no matter what you wrote" --->>> PUSH ENTER 5. waitting for answer. Probably newer ending but if enndig repeat one again else. 6. "10 rows with question no matter what you wrote" --->>> PUSH ENTER

GiteaMirror commented

2026-04-20 00:48:14 -05:00

@silentoplayz commented on GitHub (Oct 17, 2025):

https://www.lenovo.com/us/en/glossary/pebkac/

@silentoplayz commented on GitHub (Oct 17, 2025): https://www.lenovo.com/us/en/glossary/pebkac/

GiteaMirror referenced this issue

2026-04-20 00:51:16 -05:00

[GH-ISSUE #18584] issue: UI freezes when querying a model with mixed-distance in knowledge base sources #18643

GiteaMirror referenced this issue

2026-04-25 08:05:15 -05:00

[GH-ISSUE #18584] issue: UI freezes when querying a model with mixed-distance in knowledge base sources #34172

GiteaMirror referenced this issue