#GPT-5.1 Thinking : "Error in message stream" after successful output generation

1 messages · Page 1 of 1 (latest)

fallen belfryBOT
#

Reported by @pliant lion

Bug Report: GPT-5.1 Thinking : "Error in message stream" after successful output generation
`Steps to Reproduce`
  1. Open GPT-5.1 Thinking on ChatGPT Plus (web interface)
  2. Send a prompt asking the model to merge two JSON configuration files (~16k characters total, well below token limits) with specific merging rules
  3. Wait for the model to process (typically 10+ minutes of thinking)
  4. Observe that the model completes generation (visible in thinking steps)
  5. Instead of displaying the output, UI shows red error bar: "Error in message stream"
`Expected Result`

The completed merged JSON output should be displayed cleanly in the chat window without errors, since the model successfully finished generating it.

`Actual Result`

Model successfully generates the complete merged JSON (visible in thinking steps), but the UI fails at delivery and shows "Error in message stream" instead of displaying the output. Error only occurs with this JSON task, other prompts work fine. Total input is only ~16k characters (~3-4k tokens), so likely a UI/streaming bug, not a token limit issue.

`Environment`

Windows 11/Firefox

#
Additional Information

Please provide relevant details to help resolve the issue, such as:

  • ChatGPT Shared Link (if applicable).
  • Screenshots or videos demonstrating the problem.

-# ➜ Need to contact support? Visit the OpenAI Help Center.