Something significant is about to land in the AI market. OpenAI’s next frontier model, internally codenamed “Spud”, completed pre-training on March 24, 2026. CEO Sam Altman confirmed the milestone to employees the same day and said the model would be ready for release in “a few weeks.” As of today, that window is open.
The story was first reported by The Information and picked up across the AI press. Altman’s internal message, according to reporting from The Decoder, described Spud as a “very strong model” that can “really accelerate the economy.” He added: “Things are moving faster than many of us expected.”
That is not standard corporate communications. It is the kind of language Altman uses when he is trying to prepare his team for something materially different.
What Greg Brockman Said
OpenAI President Greg Brockman went further in characterising the model. He described Spud as the result of “two years of research” with a “big model feel — it’s not an incremental improvement, it’s a significant change in the way we think about model development.”
That framing matters. OpenAI has released a series of models over the past 18 months that were important but largely represented iterative progress: GPT-5, GPT-5.2, GPT-5.2-Codex, GPT-5.3, GPT-5.4. Spud, if Brockman’s description is accurate, is meant to be architecturally different rather than simply better on benchmarks.
Whether the commercial release name will be GPT-5.5 or GPT-6 remains unconfirmed. OpenAI has not announced a name. The decision, according to multiple reports, will depend on how large the capability jump actually is compared to GPT-5.4. If the gap is substantial enough, it becomes GPT-6.
Why This Is a Bigger Deal Than Usual
Several things make this model release more significant than the typical OpenAI cadence.
Resources were redirected to make it happen. OpenAI shut down Sora, its AI video generation app, on March 24, the same day Altman announced pre-training was complete. Multiple reports confirm that freeing up GPU capacity was part of the reason for the shutdown. When a company kills an existing product to clear compute for something new, that something new is a priority, not an experiment.
It is positioned as a “superapp” foundation. Reports indicate Spud could serve as the backbone for an OpenAI desktop superapp that combines ChatGPT, the Codex coding agent, and a browser product codenamed Atlas. If accurate, this would shift OpenAI from a model provider toward a vertically integrated software platform. That is a significant business model expansion.
The prediction market signal is strong. Prediction markets are assigning roughly 78 percent probability to a Spud release before April 30, 2026. These markets have tracked OpenAI releases accurately in the past. The window is narrow and the confidence is high.
Anthropic is right behind them. Anthropic revealed Claude Mythos in early April, described as its most capable model yet, though access has been limited to a small group of strategic partners. OpenAI will not want to let that narrative sit uncontested for long.
What This Means for Business
If you are currently building AI-assisted workflows on top of GPT-5.4 or planning to, the imminent arrival of Spud/GPT-5.5 or GPT-6 does not necessarily mean you should wait. Here is why.
Waiting for the next model is a trap. There will always be a better model coming. The businesses getting real operational value from AI right now are the ones that built workflows on what was available, not the ones that held off for the next version. When Spud ships, those businesses will upgrade and keep moving. Businesses that have been waiting will start at zero.
The capability jump will open new use cases, not replace current ones. If Spud really is architecturally different and significantly more capable than GPT-5.4, it will likely unlock AI applications that are not viable today: more complex multi-step agents, better long-context reasoning, and more reliable agentic behaviour. That is additive to what is already working, not a replacement for it.
The cost equation is improving, not worsening. Every major model release in recent memory has been accompanied by price reductions for previous-generation models. A GPT-6 launch would likely make GPT-5.x more affordable to run at scale. Businesses with production deployments on current models would see their costs drop.
The competitive window is still open. The fact that OpenAI is about to release a model described as “not incremental” should be a prompt to move faster on AI adoption, not slower. Your competitors are watching the same news. The companies that figure out how to deploy AI agents in their operations over the next six months will be building advantages that are hard to replicate later.
For most businesses, the right response to this news is not “let us wait and see what Spud can do.” It is “let us make sure our team has the data literacy and operational processes to take advantage of more capable AI when it arrives.”
If you want the playbook other teams are using with Claude and Codex right now, grab the free Working With Claude field guide. Download it here.