GPT-5.6 rumors are heating up with users swearing that ChatGPT has suddenly become smarter

Something looked different on ChatGPT this week, and a lot of people noticed right away.

Over at

Developer: Anshu Chemala Post a video side by side on Thursday compared the one-shot landing pages, captioning it: “Well, well, I’m one of the lucky ones who got early access to GPT-5.6 Pro.”

Developer Dobroslav Radosavljevic Published on X That everything he runs inside Codex, OpenAI’s programming agent, “looks very different from Model 5.5.” The responses below are divided between believers and people who call it a placebo.

The clearest pattern across posts is time. Conor Dart, one of the many X users amplifying the rumours, ran the test using a quick-to-use 3D browser game – with physical controls and a camera – which took just over an hour to create, versus GPT-5.5 Pro’s usual 10-minute mark.

“Not ideal, but for real-time AI game development testing, this is very impressive,” Dart wrote.

Amnesty International Insider Chitas Lua I mentioned A similar slowdown is testing a robotic simulation, and he’s also sure his results come from the new OpenAI model: “GPT 5.6 Pro continues to move (The human story 5) in a 3D test books. “Working on games once too.”

In a separate post, he noted that response times extended to 20 or 40 minutes, a pace he said had not been seen since before GPT-5.5 shipped.

Not all comparisons satisfy the popular model. Chris, one of the AI ​​experts at

“As I said before, based on a lot of reference, GPT-5.6 will be an incremental/strong improvement over GPT-5.5, not a myth-killer,” booksnoting that Fable 5 still outperforms both models in basic spaceship geometry. “My rough prediction was that it would trade blows with Fable 5 in some benchmarks, and maybe win by about half depending on the category, but it’s not clear that it would surpass it overall.”

A separate post is attributed to leaker Pankaj Kumar Detailed leaks Moving forward: The lore cut has been pushed back to December 2025, a logic effort that some testers are calling “juice value” has allegedly been raised to 960 from 768, and creating SVG and 3D designs is strong enough to beat Fable 5 in some tasks.

None of it comes from OpenAI, but the details are consistent across accounts: stronger logic, an incomplete front-end, and a release candidate called Kindle-Alpha.

An AI influencer known as Liu, citing unnamed sources, Written on the topic That the suspected model is “now being surreptitiously tested when 5.5 Pro is selected in ChatGPT,” at least for some Pro accounts, with a public launch planned for the following Thursday, June 25.

The closest thing to an OpenAI fingerprint on any of this is the memo, not the tweet. Chief Scientist Jacob Paczocki reportedly told staff that the next model is Meaningful improvement More than GPT-5.5, according to a report by Information. This still isn’t confirmation of A/B testing, a release date, or any specs floating around for the X, but it does confirm that there’s a new model brewing.

Decryption I reached out to OpenAI to ask if GPT-5.6 is being tested within ChatGPT, but the company did not respond by press time.

Why OpenAI might be in a rush

If OpenAI is rushing to roll out a new flagship model, it has reasons to be. China’s model is open source GLM-5.2 Cloud Opus 4.8 trails just one point on FrontierSWE — a benchmark that scores AI agents on multi-hour open engineering projects by dominance rate — while outright beating GPT-5.5 on the same test.

Meanwhile, Anthropic is dealing with damage he’s done himself. The company’s flagship Mythos 5 and Fable 5 models remain in tow under the US Export Control Directive It was released on June 12 due to a disputed jailbreak vulnerability, which left a gap at the top of the market that both GLM-5.2 and the default GPT-5.6 could fill.

If Anthropic CEO Dario Amodei and President Donald Trump make peace, Fable 5 will be far more powerful than any other model currently available, as the quality gap between Anthropic’s top model and OpenAI’s model is much larger than before.

There is also money on the table. It is said that OpenAI Weight of price cuts On tokens, they charge fees to developers and companies, according to Wall Street Journalin anticipation of Anthropic doing the same as the two companies prepare for dueling IPOs.

Whether any of them represent an actual release of GPT-5.6 is something only OpenAI can confirm, and the company has remained quiet through a week of leaked checkpoints and stealth testing claims. Polymarket However, traders are not waiting – prices for contracts launched between June 22 and 28 have risen by 89% this week.