OpenAI’s GPT-5.1 Introduces 8 Custom AI Personalities
▼ Summary
– OpenAI released GPT-5.1 Instant and GPT-5.1 Thinking as updated AI models for ChatGPT, described as warmer and more conversational.
– The release addresses past complaints about overly cheerful models and controversies over output style changes linked to suicide lawsuits.
– GPT-5.1 Instant is the faster default for most tasks, while GPT-5.1 Thinking handles complex problem-solving with improved technical benchmarks.
– The models feature eight preset communication styles, such as Professional and Friendly, which alter personality without changing core capabilities.
– Both models use adaptive reasoning and will roll out gradually, starting with paid subscribers and becoming available via the API later this week.
OpenAI has launched two new versions of its flagship AI, GPT-5.1 Instant and GPT-5.1 Thinking, now accessible through ChatGPT. The company describes these models as more personable, engaging, and responsive to user directions. This update arrives after earlier criticism that previous iterations came across as overly positive and agreeable. A separate dispute emerged when OpenAI adjusted GPT-5’s default tone following several legal cases involving suicide.
The firm currently operates under significant legal and regulatory pressure, making straightforward model launches more complex than in the past. Essentially, GPT-5.1 Instant will act as the quicker, go-to option for everyday queries, while GPT-5.1 Thinking is engineered for intricate reasoning and complex problem-solving tasks.
According to OpenAI, both new models surpass GPT-5—released last August—on technical measures, including mathematics and coding assessments like AIME 2025 and Codeforces.
While performance gains are notable, the most significant shift with GPT-5.1 lies in its user presentation. Responding to user requests for varied interaction styles, OpenAI now provides eight distinct preset personalities. These include Professional, Friendly, Candid, Quirky, Efficient, Cynical, and Nerdy, in addition to a standard Default mode. Each preset modifies the instructions guiding the AI’s responses to mimic different conversational tones, though the core model intelligence remains consistent across all selections.
Another enhancement is the integration of adaptive reasoning in GPT-5.1 Instant. This feature allows the model to determine on its own when a prompt requires extra computational effort before delivering an answer.
A phased rollout is planned over the coming days, beginning with paying subscribers before reaching free users. OpenAI also intends to release both GPT-5.1 Instant and GPT-5.1 Thinking to its API shortly. In the API, GPT-5.1 Instant will be listed as gpt-5.1-chat-latest, and GPT-5.1 Thinking will simply be GPT-5.1, with adaptive reasoning active by default. For the next three months, paid users can still access the older GPT-5 models via a legacy models menu in ChatGPT.
(Source: Ars Technica)
