OpenAI releases ChatGPT 5.3 Instant with fewer refusals and improved answers

OpenAI has released a new update to its ChatGPT platform, introducing the GPT-5.3 Instant model. The update focuses on improving everyday interactions by making responses more direct and reducing behaviours that users previously found overly cautious or patronising.

Unlike some earlier releases that emphasised benchmark performance, GPT-5.3 Instant concentrates on practical improvements to how the system responds to questions. The company says the model aims to provide clearer answers with less unnecessary commentary while maintaining safety standards.

GPT-5.3 Instant in ChatGPT is now rolling out to everyone.

More accurate, less cringe.https://t.co/oJpXsp9TBc

— OpenAI (@OpenAI) March 3, 2026

The release has been rolled out to all ChatGPT users from 6 March, while the previous GPT-5.2 Instant model will remain available to paid subscribers until 3 June 2026, after which it will be retired.

A shift towards more direct responses

One of the main changes in GPT-5.3 Instant is a reduction in lengthy disclaimers and refusals that previously appeared in responses. Earlier versions, particularly GPT-5.2 Instant, were often criticised for responding to sensitive or complex questions with extensive warnings or introductory text before addressing the query itself.

With the new update, OpenAI has adjusted the system to focus more directly on answering questions. The model is designed to avoid starting responses with long explanations or moral commentary, which some users felt slowed interactions and made them less helpful.

OpenAI has also modified the tone of the model’s conversational responses. Phrases that previously appeared in some answers, such as “Stop. Take a breath,” have been scaled back. The intention is to make the system sound more natural and less intrusive during conversations.

The changes aim to make ChatGPT feel more like a practical assistant rather than a system that interrupts with guidance or judgment before providing information. The overall goal is to allow users to ask questions and receive answers more quickly, while still maintaining responsible safeguards within the system.

Improvements to web search and accuracy

Another area of development in GPT-5.3 Instant is how the model uses web search results. Earlier versions of ChatGPT sometimes presented users with large groups of links or loosely connected sources that required further reading to find a clear answer.

In the latest update, the system integrates information from web searches with its own knowledge base to produce more complete responses. Instead of listing multiple links, the model aims to highlight the key information first and place supporting details within the explanation.

This approach is intended to reduce the time users spend searching through multiple sources to find relevant information. By combining search results with contextual knowledge, the system can present more focused answers directly in the conversation.

OpenAI also reports improvements in factual accuracy. According to the company, hallucination rates have decreased by up to 26.8% when the model uses web search on high-stakes topics such as medicine and law. When the model relies solely on its internal knowledge, hallucinations are reduced by around 19.7%.

These improvements are designed to make the system more reliable, particularly when dealing with subjects where factual precision is essential. The company emphasises that reducing the presence of incorrect or fabricated information remains a priority for future updates.

Rollout and remaining limitations

GPT-5.3 Instant is being rolled out gradually to all ChatGPT users starting 6 March. OpenAI confirmed that GPT-5.2 Instant will remain available for paid subscribers for a transition period before it is retired on 3 June 2026.

The company says the update reflects a broader shift in the artificial intelligence industry towards more efficient and concise digital assistants. Users increasingly expect systems to deliver clear answers quickly rather than extended explanations or conversational filler.

However, OpenAI acknowledged that the improvements are not uniform across all languages. The company noted that conversational tone in languages other than English still requires further refinement, suggesting additional updates may follow.

Despite these limitations, the release marks a step towards simplifying interactions with AI systems. By reducing unnecessary refusals and improving how information is presented, GPT-5.3 Instant aims to make everyday conversations with ChatGPT faster and more practical for users.