OpenAI Launches GPT-5.1 With Adaptive Reasoning and New Tone Controls

OpenAI has released GPT-5.1, an incremental but strategic upgrade to its GPT-5 family that pairs “adaptive reasoning” with new personalization controls designed to make interactions both more accurate and more conversational.

The company says the update includes two variants, ‘GPT-5.1 Instant’ for everyday, low-latency use and ‘GPT-5.1 Thinking’ for deeper, clearer reasoning, and that the rollout begins today with paid customers before widening to the rest of its user base.

OpenAI described GPT-5.1 Instant as “warmer, more conversational and better at following instructions,” while GPT-5.1 Thinking is positioned as an advanced reasoning engine that adapts how long it “thinks” to match task complexity, faster on simple queries, more persistent on hard ones.

The company says both variants incorporate adaptive reasoning so the model can decide when to allocate more compute to deliver a fuller answer.

Rollout, Access & APIs

The firm is prioritising Pro, Plus, Go and Business subscribers for the initial rollout, with free and logged-out users receiving access afterward; enterprise and education customers are being offered a seven-day early-access toggle.

OpenAI said the new models will be added to its API (the chat model will appear as gpt-5.1-chat-latest and the Thinking variant as gpt-5.1). Legacy GPT-5 editions will remain available for a transition period.

OpenAI chief executive Sam Altman announced the update on X, calling GPT-5.1 “a nice upgrade” and singling out improvements in instruction following and the model’s adaptive thinking behaviour. The company and Altman also highlighted easier customization of ChatGPT’s voice and behaviour through presets or manual tuning.

Personalisation & Tone Controls

A headline feature of the release is expanded personality and tone controls. Users will find refined presets, Default, Friendly, Efficient, plus newly emphasised tones such as Professional, Candid and Quirky, and can tune reply warmth, concision and even emoji frequency.

OpenAI says these settings will apply across models and that ChatGPT can suggest preference changes during a conversation when it detects a user wanting a different tone.

Benchmarks & Performance Claims

OpenAI and early reporters say GPT-5.1 shows measurable gains on technical benchmarks, notably math and coding tests such as AIME 2025 and Codeforces, improvements the company ties to the adaptive reasoning and clearer explanatory style of the Thinking variant. Independent observers note that the changes are as much about communication and stability as raw capability.

Analysts say the most consequential effects may be behavioral: fewer tone-mismatch edits, steadier coding outputs and less need to repeat or re-shape prompts for consistent results, frictions that add hidden operational cost inside teams and customer-facing systems. The update aims to reduce those clean-up cycles and make ChatGPT easier to integrate into workflows where consistent voice and fewer correction loops matter.