OpenAI has been quiet about its next major release, but the silence is breaking. In a recent appearance on the Big Technology Podcast, President Greg Brockman offered a rare glimpse into the company’s forthcoming foundation model — codenamed “Spud” internally — describing it as the most significant advancement since GPT-4.
Two Years of Research in One Model
Brockman positioned Spud not as an incremental update but as a fundamental shift in how OpenAI approaches model development. “It’s worth two years of research,” he noted, suggesting that this single model synthesizes multiple research threads the company has been pursuing. The term “big model smell” was coined during the conversation — a playful nod to the excitement of unboxing something genuinely new.
Industry speculation points to GPT-5.5 as the eventual release name, though OpenAI has not confirmed this. What is clearer is that Spud represents a fresh pretraining run, departing from the optimization-focused iterations that characterized recent updates.
What Makes Spud Different
While details remain limited, several themes emerge from Brockman’s comments and surrounding industry chatter:
New pretraining approach — Unlike recent releases that built on existing architectures, Spud appears to start from a new base, potentially incorporating advances in emergent reasoning and multimodal understanding that OpenAI has explored in research.
AGI as the North Star — Brockman explicitly framed Spud as a “major step toward AGI,” signaling that this model is designed with longer-term capabilities in mind, not just benchmark improvements.
Enhanced reasoning architecture — The emphasis on consolidating years of research suggests improvements in chain-of-thought reasoning, tool use, and multi-step problem solving — areas where current models still show limitations.
The Competitive Context
OpenAI’s move arrives amid intensifying competition. Anthropic’s Claude family continues to gain enterprise traction, Google’s Gemini suite pushes on multimodal capabilities, and open-weight models from Mistral and others challenge the closed-model paradigm. With $122 billion in funding raised, OpenAI has the resources to make Spud a definitive statement.
Early reactions from the AI research community have been optimistic, with interest particularly high around how Spud might address known weaknesses in existing models — notably hallucination rates and token limits in extended reasoning tasks.
Looking Forward
No official release date has been announced, but industry insiders suggest an unveiling within the next few weeks. Spud will likely debut in ChatGPT before API availability, following the pattern established with previous flagship releases.
If Brockman’s framing holds true, Spud could mark the beginning of a new chapter in AI capability — one where the phrase “artificial general intelligence” shifts from aspiration to observable progress.
Stay tuned to AI News for updates as more details emerge about OpenAI’s Spud model and its implications for the broader AI ecosystem.