OpenAI launches ChatGPT-5.1: CEO Sam Altman says it brings intelligence and style improvements

OpenAI is releasing GPT-5.1, overhauling its flagship model with two smarter variants—GPT-5.1 Instant and GPT-5.1 Thinking—that promise warmer conversations and sharper instruction-following. The refresh arrives three months after GPT-5's rocky August launch, when users complained the model felt cold and often fumbled tasks that older versions handled better.

GPT-5.1 Instant powers most ChatGPT interactions with snappier, more natural dialogue, while GPT-5.1 Thinking tackles complex problems with extended reasoning. Both models promises to gauge question difficulty on the fly, devoting extra processing to tough queries while firing off quick answers to simple ones.

ChatGPT-5.1 brings eight new personalities, stronger benchmarks and clearer explanations

The standout feature lets users pick from eight personality modes: Default, Professional, Friendly, Candid, Quirky, Efficient, Nerdy, and Cynical. Beyond presets, fine-tuned controls adjust how terse or warm responses feel, even how many emojis pop up. Changes kick in instantly across every chat, old and new.

According to CEO Sam Altman, it is a nice upgrade.

"GPT-5.1 is out! It's a nice upgrade. I particularly like the improvements in instruction following, and the adaptive thinking. The intelligence and style improvements are good too," Altman said in a post on X.

"Also, we've made it easier to customize ChatGPT. You can pick from presets (Default, Friendly, Efficient, Professional, Candid, or Quirky) or tune it yourself," he added.

Paid users on Pro, Plus, Go, and Business tiers get first access, with free accounts following soon after. Enterprise and Education plans receive a week-long preview window before GPT-5.1 becomes standard. Legacy GPT-5 models stick around for three months in a dropdown menu for paying subscribers.

OpenAI says the GPT 5.1 make some significant improvements on math and coding tests like AIME 2025 and Codeforces. GPT-5.1 Thinking strips away technical jargon, explaining complex concepts in plain language. The company emphasized that users wanted AI that's "not only smart, but also enjoyable to talk to," with CEO of Applications Fidji Simo noting the update balances intelligence with approachability.

Both new models will roll out to the API later this week, with gradual deployment over several days to maintain platform stability.