OpenAI Unveils Powerful New ChatGPT Updates: GPT-5 Gets Smarter and More Flexible

OpenAI is once again raising the bar for AI chat experiences with the latest updates to ChatGPT. CEO Sam Altman recently shared on X (formerly Twitter) that GPT-5 users now have more control over how the AI responds, with selectable modes such as ‘Auto,’ ‘Fast,’ and ‘Thinking.’ While Auto remains the default choice for most users, these new options allow for tailored interactions depending on the complexity or speed needed for a given task. Beyond customization, the updates also address user feedback on creativity, responsiveness, and overall model flexibility, promising a more engaging and versatile AI experience.

the Updates

OpenAI’s GPT-5 now offers selectable response modes, giving users the choice between Auto, Fast, and Thinking modes. Auto is the standard mode for most, while the Thinking mode allows for deeper, more deliberate answers. Rate limits have been set at 3,000 messages per week for GPT-5 Thinking, with additional capacity available via GPT-5 Thinking mini if the limit is reached. Impressively, the context limit for GPT-5 Thinking has been expanded to 196,000 tokens, allowing for more extensive conversations and data processing.

Additionally, the GPT-4o model has returned as a default option for all paid users, ensuring continuity for those who prefer it. OpenAI also added a “Show additional models” feature in ChatGPT’s web settings, giving subscribers access to models like o3, 4.1, and GPT-5 Thinking mini. GPT-4.5, however, remains exclusive to Pro subscribers due to high GPU costs.

In terms of personality, OpenAI plans to refine GPT-5’s responses to feel “warmer” without being as casual as GPT-4o. Users can expect future updates that allow further personalization of the AI’s conversational style. These enhancements come in response to user criticisms that GPT-5 was previously overly cautious, less creative, and restrictive compared to earlier versions. Altman reassured the community that OpenAI is committed to improving model variety, responsiveness, and usage limits to ensure the subscription fee delivers clear value.

What Undercode Says:

The latest GPT-5 updates indicate OpenAI is strategically balancing user customization with technical constraints. By introducing selectable modes, OpenAI caters to different user needs—Fast for efficiency, Thinking for depth, and Auto for everyday tasks—demonstrating a nuanced understanding of how AI is consumed. The expanded token limit of 196,000 is significant; it allows GPT-5 Thinking to handle more complex discussions, including multi-step reasoning or content creation, without losing context.

Returning GPT-4o as a default paid option is a smart move. Many users rely on familiarity and consistency, and this ensures loyal customers aren’t alienated. The “Show additional models” feature further reflects OpenAI’s intent to diversify offerings without overwhelming users. For developers and power users, these choices unlock more experimentation and customization.

The personality refinement is particularly interesting. GPT-5’s aim to feel warmer, yet less casual than GPT-4o, suggests OpenAI is working toward a balance between relatability and professionalism. This could have significant implications for enterprise adoption, where AI must be approachable but maintain credibility.

The criticism regarding creativity and restrictiveness is not trivial. While caution is essential to prevent harmful outputs, overly restrictive models can frustrate creative users. Altman’s pledge to expand model variety, increase usage limits, and improve responsiveness signals that OpenAI is listening closely to user feedback—a critical move to sustain engagement and satisfaction.

From a technical perspective, these updates also hint at OpenAI optimizing resource allocation. Offering high-GPU-cost models like GPT-4.5 only to Pro users ensures infrastructure sustainability while giving serious users access to cutting-edge performance. The incremental enhancements, rather than radical changes, suggest a roadmap that balances innovation with stability.

Overall, the GPT-5 updates are not just about adding features—they are about enhancing user control, model versatility, and personalization, all while addressing feedback from the AI community. As AI continues to embed itself in both professional and creative workflows, these upgrades position OpenAI to maintain a leading edge.

🔍 Fact Checker Results:

✅ GPT-5 now offers selectable response modes: Auto, Fast, Thinking.
✅ GPT-5 Thinking has a 196,000 token context limit and 3,000 weekly messages.
❌ GPT-4.5 is not available to all users; it remains Pro-exclusive.

📊 Prediction:

With these upgrades, ChatGPT usage is likely to grow among professional and creative users who require high-context, in-depth AI interactions. Customizable modes and personality tweaks may drive broader adoption in enterprise settings, while token expansion could make GPT-5 a preferred tool for writers, researchers, and developers needing complex multi-step reasoning. Expect OpenAI to continue iterative improvements, gradually increasing AI capabilities while maintaining user trust and subscription value.