GPT-5.1 improves ChatGPT’s intelligence and conversational style

OpenAI has announced the launch of GPT-5.1, which deploys in two versions: GPT-5.1 Instant and GPT-5.1 Thinking. The update incorporates improvements in response capability and allows models to adapt their operation according to the type of task.

GPT-5.1 Instant, the most used model in ChatGPT, integrates adaptive reasoning for the first time. This functionality allows the system to decide when it needs to process additional information before responding to complex questions, while maintaining speed in simple queries. The company has recorded significant improvements in mathematics and programming evaluations such as AIME 2025 and Codeforces.

The GPT-5.1 Thinking variant distributes processing time more dynamically than its predecessor. According to OpenAI data, in a representative distribution of ChatGPT tasks, this model responds approximately twice as fast on simpler tasks and takes twice as long on more complex ones. The responses also feature less technical terminology and undefined terms.

Regarding customization, OpenAI has updated conversational style options. The Default, Friendly, and Efficient categories remain with modifications, while Professional, Candid, and Quirky are added. These personality settings work with all models.

The company is experimentally implementing more granular controls that allow adjusting specific characteristics such as conciseness, warmth of responses, visual structure, and emoji frequency. The system can suggest updates to these preferences during conversations when it detects requests for a particular tone.

The rollout began for paid users (Pro, Plus, Go, Business) and will extend to free users. Enterprise and Edu plans will have a seven-day early access period. After that period, GPT-5.1 will become the default model.

Previous GPT-5 models will remain available for three months in the legacy models menu for paid subscribers. OpenAI will implement GPT-5.1 Instant and Thinking in its API throughout this week, with the names gpt-5.1-chat-latest and GPT-5.1 respectively.

Key points

• OpenAI launches GPT-5.1 in two versions: Instant and Thinking
• GPT-5.1 Instant incorporates adaptive reasoning for the first time
• The model decides when to process more information based on query complexity
• GPT-5.1 Thinking responds twice as fast on simple tasks and takes more time on complex ones
• Three new conversational style options are added: Professional, Candid, and Quirky
• Users will be able to adjust conciseness, warmth, visual structure, and emoji usage
• The rollout started for paid users and will reach free users
• After seven days of early access in Enterprise and Edu, GPT-5.1 will be the default model