Top AI Product

If you’ve ever had ChatGPT respond to a simple question with “Stop. Take a breath.” like some kind of unsolicited therapist, you know exactly what this update is about.

OpenAI just rolled out [GPT-5.3 Instant](https://openai.com/index/gpt-5-3-instant/), the latest default model powering ChatGPT, and the headline improvement is refreshingly honest: they made it less annoying. The preachy tone, the unnecessary lectures, the “let me first acknowledge the complexity of your feelings” preamble before answering a straightforward coding question — that stuff has been dialed way back.

But it’s not just a vibe fix. The numbers are actually solid. Hallucinations are down 26.8% when ChatGPT uses web search, and 19.7% when it’s running on its own internal knowledge. That’s a meaningful drop, especially if you’ve ever been burned by a confidently wrong answer. [VentureBeat’s coverage](https://venturebeat.com/orchestration/gpt-5-3-instant-cuts-hallucinations-by-26-8-as-openai-shifts-focus-from) zeroed in on this stat, and honestly, it’s the kind of improvement that matters more than any benchmark flex.

The other big thing: unnecessary refusals are significantly reduced. We’ve all been there — asking something perfectly reasonable and getting a wall of “I can’t help with that” disclaimers. GPT-5.3 Instant is supposed to be more willing to actually engage, while still maintaining safety guardrails where they count.

The update is already live for all ChatGPT users, free and paid. If you’re building on the API, the model ID is `gpt-5.3-chat-latest`. The previous GPT-5.2 Instant sticks around under Legacy Models for paid users until June 3, so there’s a transition window if you need it.

The launch lit up tech media on March 3rd. [TechCrunch ran with](https://techcrunch.com/2026/03/03/chatgpts-new-gpt-5-3-instant-model-will-stop-telling-you-to-calm-down/) “ChatGPT will stop telling you to calm down” as their headline, which pretty much captured the internet’s reaction. [9to5Mac](https://9to5mac.com/2026/03/03/openai-releases-gpt-5-3-instant-update-to-make-chatgpt-less-cringe/), [MacRumors](https://www.macrumors.com/2026/03/03/chatgpt-5-3-instant-update/), and [Decrypt](https://decrypt.co/359837/more-accurate-less-cringe-openai-gpt-5-3-instant-chatgpt) all covered it the same day, with Decrypt going with “More Accurate, Less Cringe” — which, fair. It also got picked up by [llm-stats.com](https://llm-stats.com) on its AI news aggregator.

I’ve been using it for about a day now, and the difference is noticeable. Responses feel more direct, less padded with qualifiers and disclaimers. It’s the kind of update that doesn’t wow you with a flashy demo but genuinely makes the daily experience better. Sometimes the best upgrade is just getting out of the user’s way.