OpenAI releases GPT-5.3 Instant in ChatGPT: Here's how it's different

OpenAI says GPT-5.3 Instant improves conversational flow, trims excessive disclaimers and delivers more reliable answers across domains

OpenAI has released GPT-5.3 Instant, an update to the default model powering ChatGPT. According to OpenAI, the update is designed to improve everyday conversations by making responses more fluid, relevant and direct, while reducing unnecessary refusals and overly cautious language. The company said the changes focus on areas that shape regular user experience, including tone, conversational flow and relevance, rather than benchmark performance alone.

Fewer unnecessary refusals and disclaimers

One of the key changes in GPT-5.3 Instant is how it handles sensitive or borderline queries. OpenAI said earlier versions, including GPT-5.2 Instant, would sometimes refuse questions that could have been answered safely or respond with what users described as overly cautious or preachy language. 

 

With GPT-5.3 Instant, OpenAI said the model “significantly reduces unnecessary refusals” and tones down overly defensive or moralising preambles before answering. According to the company, when a useful answer is appropriate, the model is designed to provide it directly without unnecessary caveats. 

In examples shared by OpenAI, GPT-5.3 Instant responded more directly to questions ranging from sports analysis to personal dating concerns. The company contrasted this with GPT-5.2 Instant, which in some cases began with broader framing before addressing the question. 

Reduced hallucinations, improved factual accuracy

OpenAI also claims improvements in factual reliability. According to the company, GPT-5.3 Instant reduces hallucination rates across multiple domains, including higher-stakes areas such as medicine, law and finance. 

In internal evaluations focused on higher-stakes topics, OpenAI said GPT-5.3 Instant reduced hallucination rates by 26.8 per cent when using web access and by 19.7 per cent when relying only on its internal knowledge, compared to prior models. In a separate evaluation based on de-identified ChatGPT conversations that users flagged for factual errors, hallucinations decreased by 22.5 per cent with web use and 9.6 per cent without web access, as per OpenAI. The company said these evaluations were designed to measure factual accuracy in scenarios where errors are more likely.

Changes to tone and writing style

OpenAI said GPT-5.3 Instant adopts a more natural and focused conversational style. The company noted that earlier versions could sometimes sound overbearing or make assumptions about user intent or emotions. 

According to OpenAI, the update reduces unnecessary proclamations and aims to keep ChatGPT’s personality more consistent across conversations. The company also said GPT-5.3 Instant performs better as a writing assistant, producing clearer and more immersive prose while maintaining coherence.

 

Availability

GPT-5.3 Instant is available to all ChatGPT users. OpenAI said the model is also available to developers through the API under the name “gpt-5.3-chat-latest”. 

OpenAI said that updates to the Thinking and Pro tiers will follow soon. GPT-5.2 Instant will remain accessible to paid users in the Legacy Models section for three months and is scheduled to be retired on June 3, 2026.