OpenAI has restored GPT-4o to the ChatGPT model picker for paid accounts and says it will give advance notice before removing models in the future.
The company made the change after pushback over GPT-5’s rollout and confirmed it alongside new speed controls for GPT-5 that let you choose Auto, Fast, or Thinking.
What’s New
GPT-4o Returns
If you are on a paid plan, GPT-4o now appears in the model picker by default.
You can also reveal additional options in Settings by turning on Show additional models, which exposes legacy models such as o3, o4-mini, and GPT-4.1 on Plus and Team, and adds GPT-4.5 on Pro.
This addresses the concern that model choices disappeared without warning during the initial GPT-5 launch.
New GPT-5 Modes
OpenAI’s mode picker lets you trade response time for reasoning depth.
CEO Sam Altman states:
“You can now choose between ‘Auto’, ‘Fast’, and ‘Thinking’ for GPT-5. Most users will want Auto, but the additional control will be useful for some people.”
Updates to ChatGPT:
You can now choose between “Auto”, “Fast”, and “Thinking” for GPT-5. Most users will want Auto, but the additional control will be useful for some people.
Rate limits are now 3,000 messages/week with GPT-5 Thinking, and then extra capacity on GPT-5 Thinking…
— Sam Altman (@sama) August 13, 2025
Higher Capacity Thinking Mode
For heavier tasks, GPT-5 Thinking supports up to 3,000 messages per week and a 196k-token context window.
After you hit the weekly cap, chats can continue with GPT-5 Thinking mini, and OpenAI notes limits may change over time.
This helps when you are reviewing long reports, technical documents, or many content assets in one session.
Personality Updates
OpenAI says it’s working on GPT-5’s default tone to feel “warmer than the current personality but not as annoying (to most users) as GPT-4o.”
The company acknowledges the need for more per-user personality controls.
How To Use
To access the extra models: Open ChatGPT, go to Settings, then General, and enable Show additional models.
That toggles the legacy list and Thinking mini alongside GPT-5. GPT-4o is already in the picker for paid users.
Looking Ahead
OpenAI promises more notice around model availability while giving you clearer controls over speed and depth.
In practice, try Fast for quick checks, keep Auto for routine chats, and use Thinking where accuracy and multi-step reasoning matter most.
If your workflows depended on 4o’s feel, bringing it back reduces disruption while OpenAI tunes GPT-5’s personality and customization.
Featured Image: Adha Ghazali/Shuterstock
