Top News

New Delhi: Right after Google releases Gemini 3.1 Flash-Lite, OpenAI has rolled out an update to its widely used ChatGPT model. The new version, called GPT 5.3 Instant, focuses on everyday conversations rather than benchmark races. The company says the model now produces more accurate answers and fewer hallucinations, a problem that has long frustrated AI users.

If you use ChatGPT often, you probably know the feeling. Sometimes the tool gives long explanations before answering a simple question. You ask for a quick explanation, and the response starts with a wall of disclaimers. OpenAI now says the latest update aims to fix exactly that.

OpenAI releases GPT 5.3 Instant update for ChatGPT

OpenAI said the new GPT 5.3 Instant model improves how ChatGPT handles daily conversations. The company focused on three main areas. Tone, accuracy, and how smoothly conversations flow.

According to the company, the update helps the model provide more relevant answers without unnecessary warnings or long introductions.

The blog post says, “GPT-5.3 Instant delivers more accurate answers, richer and better-contextualized results when searching the web, and reduces unnecessary dead ends, caveats, and overly declarative phrasing that can interrupt the flow of conversation.”

This version is already live for ChatGPT users. Developers can also access it through the API using the model name gpt-5.3-chat-latest.

Fewer refusals and fewer disclaimers

One of the biggest complaints from users has been unnecessary refusals. Earlier versions of the model sometimes rejected harmless questions or added long safety explanations before giving an answer.

OpenAI said the new model reduces those interruptions.

The company explains, “GPT-5.3 Instant significantly reduces unnecessary refusals, while toning down overly defensive or moralizing preambles before answering the question.”

In practical use, this means the AI should answer the question directly instead of adding several paragraphs about what it cannot do.

Hallucination rates drop across tests

Another key change involves factual accuracy. AI models sometimes generate incorrect facts. This issue is often called hallucination.

OpenAI says the new model shows clear improvement in internal tests.

The company reported the following results:

• Hallucinations reduced by 26.8 percent when using web results
• Hallucinations reduced by 19.7 percent when relying on internal knowledge
• In user feedback evaluations, hallucinations dropped 22.5 percent with web use
• Errors fell 9.6 percent without web access

These tests focused on high risk areas such as medicine, law, and finance, where factual mistakes can cause bigger problems.

Better web search answers

OpenAI says the model now handles internet search results more intelligently.

Earlier versions sometimes copied search results or produced long lists of links. GPT 5.3 Instant attempts to combine online information with its own reasoning.

OpenAI’s blog post states, “It more effectively balances what it finds online with its own knowledge and reasoning.”

This approach aims to produce clearer answers instead of simple summaries of search results.

Writing and conversation style updated

OpenAI has also adjusted the tone of the model. Earlier versions sometimes sounded overly dramatic or assumed too much about the user’s emotions.

The company described the issue openly in its update notes. Some responses felt awkward or overly emotional. The new version aims for a calmer and more natural conversational style.

Another improvement relates to writing tasks. OpenAI says GPT 5.3 Instant performs better when helping users write stories, articles, or creative text.

Availability and future updates

The new GPT 5.3 Instant model is now available to all ChatGPT users. Developers can start using it immediately through the API.

OpenAI confirmed that the older GPT 5.2 Instant model will remain available for three months before being retired on June 3, 2026.

The company also hinted that further updates are on the way. Improvements in language support and tone consistency remain ongoing work.