OpenAI has released GPT-5.1 Instant and GPT-5.1 Thinking, introducing major improvements in instruction handling, reasoning behavior, and conversational style. The update begins rolling out to paid ChatGPT users today, with access expanding to free users afterward.

OpenAI says GPT-5.1 responds to long-standing user feedback: people want AI that feels natural to interact with, stays on-task, and adjusts its reasoning effort based on the complexity of the prompt.

Updates to GPT-5.1 Instant

GPT-5.1 Instant, the default model for most users, now delivers a warmer, more conversational tone. OpenAI reports significantly better instruction adherence, meaning the model focuses on the exact question asked rather than drifting into unrelated explanations.

Instant also introduces adaptive reasoning, allowing the model to decide when deeper thinking is required before producing an answer. The result: faster replies for simple queries and more thorough, considered output for difficult tasks.

Enhancements in GPT-5.1 Thinking

GPT-5.1 Thinking, the advanced reasoning model, takes adaptive processing further. According to OpenAI, it now adjusts its computation time based on task difficulty, running about twice as fast for easy tasks and about twice as slow for complex prompts, compared to GPT-5 Thinking.

Responses use less jargon and fewer undefined technical terms to make the model more accessible for workplace, research, and technical communication scenarios.

Expanded Personalization and Tone Controls

OpenAI has reworked its preset personality styles to better match real-world user behavior. Updated options include:

• Default
• Friendly (formerly Listener)
• Efficient (formerly Robot)
• Professional
• Candid
• Quirky

Original options like Cynical and Nerdy remain available.

Beyond presets, OpenAI is testing fine-grained tuning controls that allow users to modify attributes such as tone warmth, brevity, scannability, and emoji usage. These adjustments now apply immediately across all active chats, rather than only new sessions.

GPT-5.1 also follows custom instructions more precisely, giving users tighter control over style, tone, and behavioral preferences.

Rollout Details

• GPT-5.1 Instant and Thinking start rolling out today for paid subscribers.
• Free and logged-out users will gain access next.
• Enterprise and Education customers receive seven days of early access before GPT-5.1 becomes the default model.
• GPT-5 Instant and GPT-5 Thinking remain available in the legacy models menu for three months.

Why It Matters

The GPT-5.1 release streamlines daily workflows by reducing the need for prompt engineering and minimizing off-intent responses. Adaptive reasoning allows the model to allocate more effort to complex tasks like data interpretation or technical explanations while speeding through simple queries.

What’s Next

OpenAI positions GPT-5.1 as a step toward highly personalized AI experiences. Updated tone styles are rolling out today, while experimental characteristic-tuning tools will reach a subset of users later this week. Further adjustments are expected as OpenAI gathers user feedback.