Reported by @karmic plover
Bug Report: GPT 5.1 is cutting off context
`Steps to Reproduce`
I don't know what you guys did, but I used to be able to submit large prompts (4K-5K lines of code) and get smart results. Things are COMPLETELY BROKEN now.
`Expected Result`
It used to work really well. It would produce code changes based on a spec and was spectacularly good.
`Actual Result`
You've obviously implemented some kind of context truncation and the model is completely losing track of what's going on. PLEASE FOR THE LOVE GOD PUT IT BACK THE WAY IT WAS. It was so good, and now... you've broken it. The model claims I didn't give it something and it admits it can't see anything from my original prompt in follow-up requests.
`Environment`
Web / Chrome