According to OpenAI’s CEO on X, ChatGPT Plus subscribers will soon have their GPT-5 rate limits doubled, they would also be able to stick with GPT-4o if they prefer, though the company is going to monitor adoption before deciding how long to keep the legacy model available.

If you noticed that GPT fall short at its response this week,you aren't imagining. Yesterday brought a noticeable dip in GPT-5’s performance after the autoswitcher failed, leaving the model feeling “way dumber” for part of the day. The issue has been fixed, and OpenAI is rolling out changes to its decision-boundary logic so the right model is picked more reliably.

Other updates include:

1. Model transparency: users will see which version answers each query.
2. UI changes to make manual “thinking” triggers easier.
3. Ongoing fine-tuning to improve GPT-5’s perceived intelligence.

Behind the scenes, the scale of the update is staggering and OpenAI claimed API traffic doubled in 24 hours. That load, combined with simultaneous feature launches, has made the rollout “a little more bumpy"

Despite the turbulence, the takeaway is clear: GPT-5 is getting faster, smarter, and more controllable and OpenAI is in no rush to abandon 4o while Plus users settle into the upgrade.