OpenAI Restores ChatGPT Model Picker
OpenAI restores ChatGPT model picker just a week after launching GPT-5, following strong feedback from users. GPT-5 was introduced as a “one size fits all” AI, using a built-in router to automatically decide the best way to answer questions. This was meant to replace the older, more complex model picker — a feature CEO Sam Altman had criticized. However, many users missed being able to choose models themselves.
Rapid Response: Model Picker Is Back
On Tuesday, Altman announced that GPT-5 now offers three modes:
Auto– Uses the router system.
Fast– Gives quicker responses.
Thinking– Slower but more detailed reasoning.
These options are available to all ChatGPT users via the restored model picker. This quick turnaround shows OpenAI’s willingness to react to user concerns.
Legacy Models Return
For paid subscribers, popular older models are back. GPT-4o is included by default, while GPT-4.1 and o3 can be enabled in settings. Their sudden removal last week sparked backlash, as many users preferred their unique tone and style. Altman assured users that OpenAI will now provide advance notice before retiring models in the future.
Personality and Customization Plans
Altman revealed that GPT-5’s personality will soon feel warmer but not as overly expressive as GPT-4o. OpenAI is also working on more customization tools, allowing each user to fine-tune the AI’s tone and behavior according to their preferences.
What This Means for Users
The return of multiple models adds complexity back to ChatGPT’s interface, showing that GPT-5’s router alone could not meet all needs. With these updates, users can enjoy both the latest GPT-5 improvements and the flexibility of choosing their favorite models.
