Issue: GPT-OSS-120B Model Generates Corrupted/Nonsensical Text with Web Search
Open
@henrypark133 is already working on this.
Since Jan 14, 2026.
bug
- Dominant language
- Rust
- Stars
- 8
- Forks
- 8
- Avg merge
- 1d 21h
- Merged PRs (30d)
- 36
Description
Problem
The openai/gpt-oss-120b model intermittently produces corrupted and nonsensical output when using web search. Responses contain JSON fragments, truncated phrases, and repetitive patterns instead of coherent answers.
Evidence:
-
Prompt = "
The U.S. President Donald Trump threatens intervention in Iran" -
/responses API return: gpt-responses.json
-
UI displays
-
From the event stream and UI screenshot:
- Output includes broken JSON: " to search news. Use web_search.": "moderate"\n}"
- Repeated corrupted patterns: " properly.earch": "moderate"\n}"
- Reasoning shows proper understanding but final output is malformed
- UI displays fragmented text: "days, to search news. Use web_search.* "moderate" and "moderate.earch*: "moderate"
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.