GPT-6 Brings Intelligent UI and Faster Answers to All ChatGPT Users

Intelligent UI: answers that become interfaces

OpenAI says it trained GPT-6 to compose responses from text, visuals and interactive elements, and to choose how they fit together based on the question. Responses can include graphics, tappable buttons, forms, charts and interactive experiences that users can work with directly in the conversation.

The format depends on the request. According to OpenAI, a comparison might work best side by side, while an explanation could be clearer as an interactive diagram. When a short text answer is most useful, ChatGPT can still give exactly that.

The company points to three main uses:

• Everyday questions: a plan can show how it fits together. In OpenAI's example, the timing for a Sunday roast sits alongside the recipe, and road trip stops can appear on a map with notes on what is worth a detour.
• Learning: visual and interactive explanations let users explore how something works, change an input to see what happens, or work through a problem step by step.
• Custom tools: users can ask ChatGPT to build a tool for the task at hand, such as a calculator to explore how savings could grow, a bill splitter for dinner with friends, or a game playable in the chat.

How it works under the hood

OpenAI describes two technical pieces. The first is a library of native, streamable components that gives each response a familiar design foundation while leaving the model room to decide how the pieces are arranged. The second is a compiler that processes the interface as the model generates it, so the interface appears progressively rather than after the whole response is finished. The company says this keeps the experience fast and familiar across web and mobile.

To teach the model design choices, OpenAI says it expanded its training methods, including evaluating the interfaces the model creates for clarity, usefulness and completeness. GPT-6 learned how to organize information clearly, when to use interactivity and when a simple text response is enough.

OpenAI acknowledges that there is still work ahead to improve the model's design judgment and expand what it can create. It describes Intelligent UI as a significant shift in how people interact with ChatGPT, with the chatbot able to combine UI, data and actions rather than fitting every question into the same conversational format.

Answering while still thinking

The second change concerns speed. With reasoning models, users could get help with harder problems but had to wait until the model finished thinking. OpenAI says ChatGPT can now interleave thinking with answering. The model was trained to take the user's waiting time into account and to build an answer across multiple partial responses, each adding useful information without unnecessary filler, while keeping the full answer as cohesive and factual as one written all at once.

OpenAI shares several results from its own internal evaluations:

• On high-value, everyday agentic tasks, GPT-6 Extra High begins answering in the same amount of time as GPT-5.6 Medium while achieving a better overall score than GPT-5.6 Extra High.
• For questions that need web search, GPT-6 Instant starts answering 44% sooner, on average, than GPT-5.6 Instant.
• On difficult problems, GPT-6 correctly addressed the key aspect of the user's question more often than GPT-5.6.

The company also says GPT-6 makes better decisions about when to look something up and more reliably finds information that supports its answer.

Safety work and what it means for users

OpenAI says GPT-6 builds on several of Astra's safety advances to better resist attempts to bypass its safety training, follow safeguards, and be clearer about what it can and cannot do, compared with GPT-5.6 Sol. Safety training was updated using lessons from real-world use to strengthen protections against high-risk misuse involving cyberattacks, biological threats and violence. In adversarial testing, the company reports stronger resistance to bypass attempts, particularly attacks that adapt across multiple turns.

For users, the release follows the first GPT-6 models, which OpenAI introduced for paid customers last month. This model is aimed at the full weekly ChatGPT audience. If OpenAI's claims hold in practice, everyday answers would arrive sooner and more often come as something users can click, adjust and use, rather than only read.