Sign In
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
New Delhi: OpenAI has started rolling out a major upgrade to its GPT-5 model line with the release of GPT-5.1, promising smarter answers and friendlier conversations. Announced on November 12, the new version brings two variants, GPT-5.1 Instant and GPT-5.1 Thinking, aimed at different kinds of users and tasks, with updated response styles and more personalisation options.
Both models are now available to ChatGPT Pro, Plus, Go and Business users. Those on Enterprise and Education plans will get early access through a toggle before GPT-5.1 becomes the default. Free and logged-out users will also see the rollout soon, but in phases.
OpenAI says GPT-5.1 Instant remains the most-used model and now feels more “playful, clear and useful” while also being better at following instructions.
“It often surprises people with its playfulness while remaining clear and useful,” the company said in its official blog.
The model now uses “adaptive reasoning” to decide how much effort to put into a query. This means it thinks harder before answering complicated questions, while still giving fast replies for simpler ones. In internal tests, the Instant model showed gains in math and coding tasks like AIME 2025 and Codeforces.
Another notable upgrade is improved instruction following. That includes handling requests like “write exactly six words” with much better accuracy.
For users handling complex queries, like breaking down code, technical tasks or deep reasoning, GPT-5.1 Thinking is the upgraded option. The model adjusts its thinking time depending on the complexity of the question.
“It now adapts its thinking time more precisely to the question, spending more time on complex problems while responding more quickly to simpler ones,” OpenAI said.
This allows the model to conserve tokens and avoid long wait times for easy tasks. At the same time, for heavy-duty prompts, it now delivers more clarity with less jargon.
“GPT-5.1 Thinking’s responses are also clearer, with less jargon and fewer undefined terms,” the blog said, adding that this makes it more approachable for technical work and explanation-heavy queries.
OpenAI is also expanding how users can customise ChatGPT’s responses.
Earlier, users could only choose between tone presets like “Default,” “Friendly,” or “Efficient.” Now, new options like “Professional,” “Candid,” and “Quirky” have been added. Two older styles, “Cynical” and “Nerdy,” remain available.
“Tailoring its tone to what sounds right for you should feel effortless,” OpenAI said, explaining that users now have more intuitive controls over how the model sounds.
The company is also testing fine-grained settings like:
The new preferences take effect immediately across all chats, including those already underway, which wasn’t the case earlier.
The launch of GPT-5 earlier this year had its share of bumps. Many early users weren’t convinced it outperformed GPT-4 in everyday writing or reasoning.
Sam Altman, OpenAI’s CEO, later admitted that issues with the router, the system that decides which model to use for each query, may have caused these problems.
In this version, the “GPT-5.1 Auto” router continues to steer questions to either Instant or Thinking depending on the nature of the request.
The legacy GPT-5 models will remain accessible via dropdown for paid users for three months, allowing time to compare versions. The sunset period won’t affect other older models like GPT-4o.
Developers will soon get access to the updated models through OpenAI’s API. GPT-5.1 Instant will be listed as gpt-5.1-chat-latest, while GPT-5.1 Thinking will show up as gpt-5.1, both featuring adaptive reasoning capabilities.
The company has also updated its system card to explain its safety approach for GPT-5.1 and promised more improvements down the line.
As the rollout begins, users can expect warmer chats, smarter replies, and new ways to make ChatGPT feel more personal, whether they’re solving math, writing code, or just having a casual conversation.
