I noticed some issues with the GPT UI and models: even when I select 5.4 Thinking mode, it still defaults to using 5.3 Instant. Additionally, 5.3 Instant often fails to understand my input clearly, and frequently just repeats what I said without adding useful responses.
Hey - quick note in case this helps.
If you’re using Auto, ChatGPT may switch between GPT-5.3 Instant and GPT-5.4 Thinking depending on the task. More complex prompts can trigger GPT-5.4 Thinking. Also worth checking the usage limits mentioned in the help doc.
If you’re not using Auto and the model is still switching unexpectedly, could you share if you’re seeing any error message when changing models, a screen recording of the behavior and whether the same thing happens in a new conversation? That usually helps tell if it’s conversation-specific or something broader.
Also on the point about 5.3 Instant repeating your input, could you share an example prompt and what response you received? That helps tell whether it’s prompt-specific or something broader.
If you’re able to capture those details, feel free to send them to support@openai.com
and we’ll be happy to take a closer look.
– Ruth
2 Likes
This topic was automatically closed after 24 hours. New replies are no longer allowed.