OpenAI is rolling out a wave of new features and platform updates to ChatGPT as part of a broader effort to improve user interaction, expand educational use cases, and lay the groundwork for the next phase in AI model development. These changes are being released gradually and are expected to shape both how users experience the chatbot and how developers engage with OpenAI’s upcoming tools.
ChatGPT Adds Personality Options Like Robot, Cynic, Listener, and Sage
A new update, as spotted by Tibor Blaho, within ChatGPT allows users to change how the assistant responds by choosing from multiple personality types. This personalization tool gives people more control over the chatbot's tone and communication style. Instead of only interacting in its usual friendly and adaptive way, the assistant can now behave in different modes, depending on what the user selects.
For instance, the “Robot” setting responds in a direct and no-nonsense manner. The “Cynic” setting introduces a more sarcastic and critical tone. There's also a “Listener” mode that focuses on being more thoughtful and supportive, while the “Sage” style leans into being knowledge-driven and enthusiastic. These options are found under the “Customize ChatGPT” menu, but the update may not yet appear for every user.
New Study Together Mode Offers Guided Learning and Homework Help
Another major addition is the “Study Together” feature, which is designed to help users understand complex topics through structured explanations. Unlike basic one-answer prompts, this mode takes a more gradual approach to learning, offering step-by-step breakdowns of concepts and encouraging users to explore questions more deeply.
The feature includes tools that support exam preparation and topic-specific quizzes, making it useful for students and anyone who wants to review material in an organized way. While the underlying model doesn’t change, the format of responses is tailored to educational use, aiming to be more interactive than standard responses. This is being rolled out incrementally across the platform.
Sora 2 in Development as OpenAI Prepares a Return to Generative Video
OpenAI is also working on a new generative video model referred to internally as “Sora 2.” The original Sora tool launched in December 2024 and gained attention for its ability to create video content using AI prompts. However, it has remained largely unchanged since its release and faded from the spotlight.
Recent server updates suggest that development on its successor is underway, though no official launch date has been shared. Sora 2 appears intended to compete directly with Google's Veo 3, which is now free for students using the Gemini AI Pro plan. OpenAI’s response could see the company re-enter the video creation space with updated capabilities, though further details are expected in the weeks ahead.
DevDay 2025 May Reveal GPT-5 or Preview Advanced Features
OpenAI’s annual developer event, DevDay, is set for October 6 in San Francisco, and it may include updates on GPT-5, a model many have been waiting to see. So far, no formal announcement confirms that GPT-5 will debut at the event, but references to an internal model named “GPT-5-reasoning-alpha” have surfaced, hinting that testing is ongoing.
Development of GPT-5 has faced delays tied to computing power constraints, especially shortages of GPUs needed to run large-scale models. Earlier this year, OpenAI launched GPT-4.5 but limited access due to infrastructure limits. As a result, many users continue to rely on GPT-4o, which remains the default model for most public interactions.
In the meantime, OpenAI has introduced two new research-grade models, o3-deep-research and o4-mini-deep-research, which serve more specialized roles rather than general use. The DevDay keynote is expected to provide insights into future releases, platform direction, and ongoing challenges related to model deployment and capacity.
Read next: Trump’s AI Policy Takes Aim at Bias, Shifts Focus to Infrastructure and Global Competition
