OpenAI is rolling out GPT-5.1, an upgrade to the GPT-5 family that makes ChatGPT sharper and more conversational, starting with paying customers before reaching free users. The release introduces two key variants, GPT-5.1 Instant and GPT-5.1 Thinking, combining higher intelligence, improved instruction following and a friendlier tone. Instant remains the high volume model but now uses adaptive reasoning, deciding when to think longer on hard prompts and lifting math and coding scores on exams like AIME 2025 and Codeforces while staying quick for chat.
👍 https://t.co/nK4qoC5tUT pic.twitter.com/Yp2wTO6qJD
— roon (@tszzl) November 12, 2025
Thinking becomes the top reasoning tier, varying its deliberation per question so simple tasks answer faster and complex work gets deeper analysis with clearer language. GPT-5.1 Auto continues routing queries between models, GPT-5.1 Instant and Thinking reach the API this week as gpt-5.1-chat-latest and GPT-5.1, and legacy GPT-5 stays selectable for about three months. At the same time, ChatGPT gains stronger personalization controls so users can pick tones like Default, Professional, Friendly, Candid, Quirky or Efficient and have that style apply across all chats.
We’re also experimenting with more personalization controls, like tuning warmth ☀️ and emoji frequency 🤩
— OpenAI (@OpenAI) November 12, 2025
Rolling out to a small group of users as we test and learn.
An experimental settings panel lets a subset of users tune brevity, warmth and emoji use, while updates now apply instantly to ongoing as well as new conversations. Combined with the new tone controls, the move pushes ChatGPT toward a configurable assistant for workers, students and large organizations, even as early power users on forums debate whether it feels like a true model change or mainly routing and UX tweaks.
