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
Currently, tasks are not enforced. A model can just terminate on its own before all tasks are completed.
As this discussion shows, models sometimes prematurely emit their stop tokens and the continuation of the conversation by the user despite that can result in the model adopting the pattern of early termination as a feature.
Desired Solution you'd like
Making tasks enforceable could avoid the failure mode of models randomly emitting their stop tokens before having completed everything that was asked of them.
Refuse to let the model give a final answer until it has completed all its tasks.
Alternatives Considered
Filters/pipes as entry points for LLM judges
AFAIK filters don't work here, but pipes might.
Additional Context
Example of model terminating on its own despite pending tasks:
Originally created by @Material-Scientist on GitHub (May 11, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/24580
### 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
Currently, tasks are not enforced. A model can just terminate on its own before all tasks are completed.
As this [discussion](https://github.com/open-webui/open-webui/discussions/24309#discussioncomment-16880219) shows, models sometimes prematurely emit their stop tokens and the continuation of the conversation by the user despite that can result in the model adopting the pattern of early termination as a feature.
### Desired Solution you'd like
Making tasks enforceable could avoid the failure mode of models randomly emitting their stop tokens before having completed everything that was asked of them.
Refuse to let the model give a final answer until it has completed all its tasks.
### Alternatives Considered
Filters/pipes as entry points for LLM judges
AFAIK filters don't work here, but pipes might.
### Additional Context
Example of model terminating on its own despite pending tasks:
<img width="1344" height="1892" alt="Image" src="https://github.com/user-attachments/assets/bf180784-9981-4580-9f2d-0b679f001d4d" />
@owui-terminator[bot] commented on GitHub (May 11, 2026):
🔍Related Issues Found
I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:
🟣#17058issue: Response cannot be stopped after the tool is called This is about responses continuing or not being cleanly stopped after a tool call, which is adjacent to the new issue’s concern about models ending a task too early or not respecting task boundaries. by EntropyYue · bug, confirmed issue
🟣#19744issue: message token cutoff not leaving enough space for completion Both issues involve generation stopping/cutoff behavior and the model failing to leave enough room or continue to a proper completion, making it the closest token-limit/termination-related match. by NedasValantic · bug
🟣#20816issue: Cancel event sent for long-running tasks, no output retrieved This discusses task cancellation and long-running task handling in the same task pipeline, which is related to enforcing task completion and preventing premature termination/cancellation. by moooV252 · bug
💡 If your issue is a duplicate, please close it and add any additional details to the existing issue instead.
This comment was generated automatically. React with 👍 if helpful, 👎 if not.
<!-- gh-comment-id:4421730080 -->
@owui-terminator[bot] commented on GitHub (May 11, 2026):
<!-- terminator-bot:related-issues-reply -->
🔍 **Related Issues Found**
I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:
1. 🟣 [#17058](https://github.com/open-webui/open-webui/issues/17058) **issue: Response cannot be stopped after the tool is called**
*This is about responses continuing or not being cleanly stopped after a tool call, which is adjacent to the new issue’s concern about models ending a task too early or not respecting task boundaries.*
*by EntropyYue · `bug`, `confirmed issue`*
2. 🟣 [#19744](https://github.com/open-webui/open-webui/issues/19744) **issue: message token cutoff not leaving enough space for completion**
*Both issues involve generation stopping/cutoff behavior and the model failing to leave enough room or continue to a proper completion, making it the closest token-limit/termination-related match.*
*by NedasValantic · `bug`*
3. 🟣 [#20816](https://github.com/open-webui/open-webui/issues/20816) **issue: Cancel event sent for long-running tasks, no output retrieved**
*This discusses task cancellation and long-running task handling in the same task pipeline, which is related to enforcing task completion and preventing premature termination/cancellation.*
*by moooV252 · `bug`*
---
💡 If your issue is a duplicate, please close it and add any additional details to the existing issue instead.
*This comment was generated automatically.* React with 👍 if helpful, 👎 if not.
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 @Material-Scientist on GitHub (May 11, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/24580
Check Existing Issues
Verify Feature Scope
Problem Description
Currently, tasks are not enforced. A model can just terminate on its own before all tasks are completed.
As this discussion shows, models sometimes prematurely emit their stop tokens and the continuation of the conversation by the user despite that can result in the model adopting the pattern of early termination as a feature.
Desired Solution you'd like
Making tasks enforceable could avoid the failure mode of models randomly emitting their stop tokens before having completed everything that was asked of them.
Refuse to let the model give a final answer until it has completed all its tasks.
Alternatives Considered
Filters/pipes as entry points for LLM judges
AFAIK filters don't work here, but pipes might.
Additional Context
Example of model terminating on its own despite pending tasks:

@owui-terminator[bot] commented on GitHub (May 11, 2026):
🔍 Related Issues Found
I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:
🟣 #17058 issue: Response cannot be stopped after the tool is called
This is about responses continuing or not being cleanly stopped after a tool call, which is adjacent to the new issue’s concern about models ending a task too early or not respecting task boundaries.
by EntropyYue ·
bug,confirmed issue🟣 #19744 issue: message token cutoff not leaving enough space for completion
Both issues involve generation stopping/cutoff behavior and the model failing to leave enough room or continue to a proper completion, making it the closest token-limit/termination-related match.
by NedasValantic ·
bug🟣 #20816 issue: Cancel event sent for long-running tasks, no output retrieved
This discusses task cancellation and long-running task handling in the same task pipeline, which is related to enforcing task completion and preventing premature termination/cancellation.
by moooV252 ·
bug💡 If your issue is a duplicate, please close it and add any additional details to the existing issue instead.
This comment was generated automatically. React with 👍 if helpful, 👎 if not.