TL;DR

OpenAI has launched GPT-5.1 Instant and GPT-5.1 Thinking, upgrading the GPT-5 series with warmer default tone and improved conversational style. GPT-5.1 Instant features adaptive reasoning that determines when to think before responding, whilst GPT-5.1 Thinking adapts processing time more precisely to question complexity. The release includes expanded tone customisation options—Professional, Candid, and Quirky presets join existing choices—and experimental controls for tuning response characteristics like conciseness and emoji frequency.

Opening

OpenAI’s GPT-5.1 release addresses user feedback that capable AI should be both intelligent and enjoyable to interact with. The upgrade introduces two model variants rolling out to paid users first, before broader availability to free and logged-out users.

Context: Intelligence Meets Personality

GPT-5.1 Instant, ChatGPT’s most-used model, now defaults to warmer and more conversational responses. Early testing indicates users often find the playfulness surprising whilst maintaining clarity and usefulness. The model includes adaptive reasoning capability, automatically deciding when to think before responding to challenging questions—reflected in significant improvements on mathematics and coding evaluations including AIME 2025 and Codeforces.

GPT-5.1 Thinking adapts processing time more dynamically than its predecessor. On representative ChatGPT tasks, the model responds roughly twice as fast on simple queries and takes twice as long on complex problems. Responses include less jargon and fewer undefined terms, making advanced reasoning more approachable for workplace tasks and technical concept explanation.

The release introduces refined tone presets: Default, Friendly (formerly Listener), and Efficient (formerly Robot) remain with updates, whilst Professional, Candid, and Quirky join the options. Original Cynical and Nerdy presets remain unchanged. These personality settings apply across all models.

For users requiring finer control, OpenAI is experimenting with characteristic tuning—adjusting conciseness, warmth, scannability, and emoji frequency directly from personalization settings. ChatGPT can proactively offer to update preferences during conversations when detecting requests for specific tone or style.

Looking Forward

GPT-5.1 Auto continues routing queries to optimal model variants, meaning most users won’t manually select between models. Personalization settings now affect all chats immediately, including ongoing conversations, rather than applying only to new threads.

GPT-5 (Instant and Thinking) will remain available via legacy models dropdown for paid subscribers for three months, allowing gradual comparison and adaptation. The sunset period provides space for evaluation and feedback whilst OpenAI continues innovating frontier models. Both GPT-5.1 variants will join the API later this week, with Instant as gpt-5.1-chat-latest and Thinking as GPT-5.1, both featuring adaptive reasoning.

Source: OpenAI