In brief
- Two models debut: GPT-5.1 Instant emphasizes warmth and conversational ease; GPT-5.1 Thinking focuses on efficiency and depth.
- Adaptive reasoning: Both models now gauge question complexity before responding, boosting math and coding performance.
- More control: Users can fine-tune personality traits and tone, with rollout starting today for paid ChatGPT tiers.
OpenAI today announced the release of GPT-5.1, an iterative upgrade to its foundational large language model series. The update, which is now rolling out to users, features two enhanced models: GPT-5.1 Instant and GPT-5.1 Thinking, focusing simultaneously on increased capability and improved conversational warmth.
The company said the update aims to address user feedback that AI should be not only intelligent but also enjoyable to interact with, positioning the new models as “smarter, more conversational ChatGPT.”
Instant vs. Thinking
The GPT-5.1 release is split across the two primary use cases within ChatGPT:
1. GPT-5.1 Instant
Designated as ChatGPT’s most-used model, GPT-5.1 Instant has been optimized to be “warmer by default and more conversational.” Key improvements include:
- Adaptive reasoning: For the first time, this widely used model can perform “adaptive reasoning,” allowing it to internally assess the complexity of a question and decide whether to engage in deeper processing before generating a response.
- Performance: This adaptive capability has led to demonstrable improvements in high-stakes reasoning evaluations, including math and coding benchmarks such as AIME 2025 and Codeforces.
- Instruction following: The model is also cited as being significantly better at reliably adhering to custom instructions and requested formats.
2. GPT-5.1 Thinking
The upgrade to the advanced reasoning model, GPT-5.1 Thinking, centers on efficiency and clarity.
- Dynamic processing: The model now adapts its thinking time dynamically to the input. It spends considerably less time on simple tasks—making it roughly twice as fast on the quickest queries—while dedicating substantially more time to complex problems, enabling more persistent and thorough answers.
- Clarity: OpenAI notes that GPT-5.1 Thinking’s responses are clearer, utilizing “less jargon and fewer undefined terms,” making its high capabilities more accessible to non-technical users.
Customization and rollout schedule
In parallel with the model upgrades, OpenAI is enhancing user control over the AI’s personality and tone. New preset options for conversational style—including Professional, Candid, and Quirky—are being introduced, refining the previous tone controls. The company also announced an experiment to allow users to tune specific characteristics, such as conciseness and emoji frequency, directly from personalization settings.
The rollout of GPT-5.1 Instant and GPT-5.1 Thinking begins today for paid subscribers (Pro, Plus, Go, Business) before expanding to free users. Enterprise and Education users will have a seven-day early-access toggle before the new models become the sole default. An API integration for developers is slated to follow later this week.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
Source link
