OpenAI has unveiled two new models in its GPT-5 series, named GPT-5.1 Instant and GPT-5.1 Thinking, both now being introduced in ChatGPT. These models aim to improve the reliability and personalisation of chat-based interactions for users in both professional and casual settings.

Model enhancements

GPT-5.1 Instant is positioned as the most frequently used model within ChatGPT. According to OpenAI, it offers a more conversational and warmer interaction style while displaying improved accuracy in following instructions. The model now utilises adaptive reasoning, specifically designed to decide when more deliberation is necessary for complex queries. This approach is intended to result in more precise and thorough responses to challenging questions, with ongoing quick reply times for simpler requests.

The company states that updates to GPT-5.1 Instant have led to better outcomes on quantitative evaluations such as the American Invitational Mathematics Examination (AIME) 2025 and programming competitions like Codeforces, marking an improvement over prior versions.

GPT-5.1 Thinking, the series’ advanced reasoning model, has been updated to be more efficient and user-friendly. It can now better adjust its processing time depending on the complexity of a task. The model provides clearer explanations with less technical jargon and is designed to deliver responses that are easier for users to understand.

Default routing

With the introduction of these upgrades, ChatGPT will continue to use its internal GPT-5.1 Auto routing feature. This system automatically selects the model best suited for each user query, removing the need for users to manually choose between Instant or Thinking for most tasks.

Users will still have access to legacy GPT-5 models for a transitional period of three months, allowing for comparisons and adjustments. The company plans for GPT-5.1 to eventually become the sole default model once the rollout is complete. GPT-5.1 Pro, an updated version for professional users, will be introduced in the near future. Both new models are expected to be made available via OpenAI’s API shortly.

Personalisation options

OpenAI is further expanding ChatGPT’s personalisation features. New and refined preset personality options offer broader choices, including Default, Friendly, Efficient, Professional, Candid, and Quirky. The original Cynical and Nerdy presets remain available. The changes are based on user behaviour and preferences, aiming to improve the ease with which users can tailor the model’s conversational style to their needs.

In addition, OpenAI is experimenting with advanced tuning settings so users can further customise ChatGPT’s characteristics. These allow adjustment of factors such as response conciseness, warmth, scan-ability, and emoji use frequency. ChatGPT may also actively prompt users to update these preferences during conversation when it detects specific requests for tone or style adjustments.

Personalisation changes now take effect across all ongoing and future conversations, rather than requiring new sessions for updates to apply. This aims to offer a more consistent user experience as preferences are changed.

Graduated rollout

The staggered introduction of GPT-5.1 Instant and Thinking will commence with paid users, including ChatGPT Plus, Go, Business, and Pro subscribers, before extending to free and logged-out users. Customers on enterprise and educational plans will have an initial period of early access. The company’s approach includes an evaluation phase to maintain service stability during uptake.

“What you will notice is that answers across GPT-5.1 feel both smarter and more natural in tone,” said OpenAI.