TLDR: OpenAI has rolled out GPT-5.1, a significant update to its ChatGPT platform, introducing ‘Instant’ and ‘Thinking’ models designed for more natural, empathetic, and instruction-following conversations. This release addresses previous criticisms of GPT-5’s stiffness, offering user customization options and adaptive reasoning, alongside new safety assessments for mental health and emotional reliance.
OpenAI has launched a substantial update to its flagship ChatGPT platform, introducing GPT-5.1, a refinement of its GPT-5 series aimed at delivering ‘sharper reasoning, improved instruction following, and a more natural conversational tone.’ The update, which began rolling out on November 12, 2025, features two enhanced models: GPT-5.1 Instant and GPT-5.1 Thinking. These models are specifically engineered to make interactions feel ‘warmer, more intelligent, and better at following instructions,’ according to OpenAI.
The release comes after a mixed reception to the initial GPT-5 earlier this year, where users lauded its power but often criticized its rigid and uneven communication style. OpenAI acknowledged this feedback, stating, ‘We’ve clearly heard from users that a great AI not only needs to be smart but also make chatting with it a pleasant experience.’ The company’s announcement emphasized that ‘GPT-5.1 improves meaningfully on both intelligence and communication style.’
OpenAI CEO Sam Altman confirmed the launch on X, describing GPT-5.1 as ‘a nice upgrade.’ He highlighted improvements in instruction following and ‘adaptive thinking,’ along with general intelligence and style enhancements. A key feature is the expanded customization options, allowing users to select from preset personalities like ‘Default, Friendly, Efficient, Professional, Candid, or Quirky,’ or to fine-tune the AI’s chatting style themselves. This directly addresses earlier criticisms regarding GPT-5’s lack of personable interaction compared to its predecessor, GPT-4o.
GPT-5.1 Instant has been re-engineered for a more approachable and emotionally aware demeanor. It dynamically adjusts its reasoning depth, providing quick responses to simple queries while taking more time to ‘think’ through complex ones. Early demonstrations suggest a more empathetic and playful tone. Similarly, GPT-5.1 Thinking, designed for complex problem-solving, now offers explanations that are ‘clearer, with less jargon and fewer undefined terms,’ and adopts a noticeably warmer tone, softening its previously logic-driven approach. OpenAI noted significant improvements in professional evaluation test sets for mathematics (AIME 2025) and programming (Codeforces).
Beyond performance, OpenAI has also introduced new human-oriented dimensions to its model’s security assessment, particularly in response to recent concerns. These include evaluating how the AI responds to signs of mental health issues like isolation, delusion, or mania, and assessing whether its responses might encourage unhealthy dependence or emotional attachment from users. This signifies a broader commitment to user well-being and responsible AI development.
Also Read:
- OpenAI Explores Consumer Health Sector with AI-Powered Tools, Report Suggests
- OpenAI Bolsters Global Startup Outreach with New Solutions Architect Roles in India and South Korea
The update is being gradually pushed to paid ChatGPT users (Pro, Plus, Go, Business) starting November 12th, with free and unregistered users receiving it subsequently. Enterprise and education versions will have a 7-day early-bird switch option before a uniform upgrade. API developers will also gain access to GPT-5.1 Instant and GPT-5.1 Thinking.
