TLDRs:
- OpenAI launches GPT-5 modes in ChatGPT, allowing users to choose Auto, Fast, or Thinking modes.
- GPT-5 Thinking mode supports up to 196,000 tokens with a weekly 3,000-message limit.
- CEO Sam Altman promises future updates for AI personality customization in ChatGPT.
- OpenAI staff awarded selective bonuses ahead of GPT-5 launch to retain top AI talent.
OpenAI has rolled out new modes for GPT-5 in ChatGPT, giving users unprecedented control over the AI’s performance.
CEO Sam Altman announced on Wednesday that users can now toggle between “Auto,” “Fast,” and “Thinking” modes, enhancing flexibility depending on the task at hand. The move aims to balance speed, depth, and contextual understanding while making the AI experience more customizable.
Paid ChatGPT subscribers now access the 4o model by default and can use a new toggle to explore additional options, including o3, 4.1, and the GPT-5 Thinking mini model. The full GPT-5 Thinking mode carries a weekly message cap of 3,000, after which users may continue via the mini version.
Notably, this mode offers a massive context limit of 196,000 tokens, positioning it competitively against other advanced AI platforms. However, GPT-4.5 remains restricted to Pro users due to high GPU costs, highlighting ongoing compute challenges for AI companies.
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
AI Personality Gets a Warm Upgrade
Altman also revealed that OpenAI is refining GPT-5’s personality to be “warmer but not as annoying,” addressing long-standing user feedback from GPT-4. Studies suggest AI personality significantly influences satisfaction, trust, and engagement.
By focusing on emotional connection alongside technical accuracy, OpenAI signals that successful AI interaction now depends as much on relatability as on raw capability.
The company is exploring per-user customization, responding to growing demand for personalized AI experiences.
Current research indicates that 71% of consumers expect tailored content, and organizations prioritizing personalization report higher engagement and revenue. By letting users shape the AI’s tone and responsiveness, OpenAI hopes to deliver a more human-like experience without compromising professional reliability.
Rewards for AI Talent
Ahead of GPT-5’s launch, OpenAI awarded select staff generous bonuses to retain critical talent amid intense industry competition.
Around a third of employees, mainly researchers and engineers, received bonuses distributed over two years, with top researchers earning mid single-digit millions.
While this move reflects the extraordinary valuation of AI expertise today, selective awards carry risks, including potential morale issues among non-recipient staff.
Looking Ahead
OpenAI also plans to expand user access to Pro-level queries for Plus users once compute capacity allows. This strategic approach demonstrates a focus on balancing accessibility, performance, and innovation while maintaining high-quality interactions.
The launch of GPT-5 modes represents a notable evolution in AI usability. By combining performance choices, high-context processing, and the promise of customizable personalities, OpenAI is signaling that future AI success will hinge as much on user experience as on technical prowess.
