While using GPT-5.6 Sol, I’ve noticed a significant drop in output quality. When I asked the model which version it was running, it claimed to be 5.5 mini, even though I had 5.6 Sol selected. Is anyone else experiencing this issue, or does anyone know a fix?
I’m affected by this issue as well. It appears that several users are experiencing similar problems. Sharing reports in the main thread may help clarify the situation and assist with further investigation.
same here Proof:
{ "metadata": {
"model_slug": "gpt-5-6-thinking",
"default_model_slug": "gpt-5-6-thinking",
"resolved_model_slug": "gpt-5-5-mini" } }
Hi everyone, thanks for flagging this and sharing the details.
Are you still seeing this mismatch? If so, please add one recent affected conversation URL, the date and time including your timezone, and redacted screenshots showing the selected model and the different model name or response metadata to your existing private Support request. If you don’t have one, please email support@openai.com.
Please keep conversation links and diagnostic details in that private request.
~ Ziad