Bug Report: Model switch causes assistant context inconsistency (previous assistant replies not preserved)
`Steps to Reproduce`
1.Start a conversation in ChatGPT.
2.Ask several questions and receive responses using the default model(ChatGPT 5.2).
3.Use the “regenerate” option and switch to a different model (ChatGPT 4o) .
4.Observe the new response after the model switch.
`Expected Result`
The newly selected model should receive the full conversation history, including previous assistant responses, and behave consistently with the existing context.
`Actual Result`
After switching models, the newly selected model appears to lack visibility into previous assistant responses generated by the default model. Instead of continuing from the existing context, the new model behaves as if multiple earlier user questions had never been answered, and produces a single aggregated response that re-answers many previously resolved topics in one message.
`Environment`
Platform: ChatGPT Web. Version: The test versions are ChatGPT 5.2 and ChatGPT 4o. Browser: Chrome.
silent lintelBOT