Update 24/04/26 – 10:05 am (IST): The rumors were completely accurate, and OpenAI has now officially launched GPT-5.5 for Plus, Pro, Business, and Enterprise users. This confirms the earlier hints from Sam Altman about a Thursday rollout following the unexpected Codex leak.
The new model is designed to act more like a highly independent digital worker that can handle messy and complex computer tasks on its own. It can browse the web, write code, and navigate different software tools to finish a project without needing you to guide it through every single step.
Original article published on April 22, 2026, follows:
GPT-5.5 seems to have briefly surfaced in OpenAI’s Codex platform a few hours ago alongside several unreleased models, right as CEO Sam Altman heavily hinted at a Thursday release.
Apparently, the internal model picker accidentally went live for some Codex users on Pro accounts. People clicking the dropdown saw an unexpected list of names. The new lineup included GPT-5.5, oai-2.1, something called arcanine, and multiple glacier-alpha checkpoints.
Here’s a screenshot shared by @marmaduke091 for reference:
The access disappeared quickly. But it was not fast enough to stop developers from testing the unreleased flagship.
One developer reported that GPT-5.5 fixed a problem in three minutes that had held them up for four hours. Others who squeezed in a prompt said the model felt much faster and significantly more token-efficient.
It seems to be a clear step up for front-end development tasks. Another user fired off a prompt to design and code a high-end, ultra-modern HTML and Tailwind CSS landing page before OpenAI pulled the plug. They noted the new model looks great at design now.
The sudden leak lines up perfectly with ongoing industry drama. Anthropic just angered its user base by stripping Claude Code out of its Pro plan for some. Developers are actively complaining on social media and threatening to cancel subscriptions.
Altman clearly saw the opening. He jumped into a furious thread on X to tell frustrated developers to “Come to the light side.”
When a user replied that they would happily switch and buy a subscription immediately if OpenAI launched GPT-5.5 or GPT-6 on Thursday, Altman responded with a salute emoji.
Some are even speculating that the Codex leak was a deliberate tease to build hype.
That said, the company has already been making moves. OpenAI just officially announced the ChatGPT Images 2.0 update, which we spotted rolling out to users yesterday. And earlier chatter already hinted at a Thursday release this week. So Altman’s salut all but confirms the same.
We’ll keep an eye out for any further developments and will update the article accordingly.
Featured image generated with AI



