OpenAI is preparing to release a new major AI model internally codenamed “Spud,” which has completed pre-training and is expected to launch soon, possibly later this month or in April. This model is anticipated to be a significant advancement, potentially accelerating the economy, and may represent a shift towards a super app that integrates various OpenAI tools like ChatGPT, Codex, and Atlas into one unified product. Alongside this, OpenAI is restructuring its leadership, with CEO Sam Altman stepping back from direct safety oversight to focus on infrastructure, fundraising, and scaling operations, signaling a strategic pivot towards building the necessary hardware and supply chains to support large-scale AI deployment.
As part of this refocus, OpenAI is discontinuing its Sora project, which was consuming substantial compute resources. Sora was linked to a high-profile but now-defunct video generation deal with Disney, which reportedly never finalized. The Sora team is pivoting towards long-term research on world simulation and robotics, aiming to develop systems that deeply understand and simulate environments, similar to efforts by Google DeepMind and Nvidia. This shift underscores OpenAI’s commitment to prioritizing foundational AI capabilities and AGI deployment, as reflected in recent organizational changes and the introduction of AGI as a formal product category in their organizational chart.
Meanwhile, OpenAI faces strong competition from Anthropic, which is rapidly advancing in enterprise AI tools and coding assistants. OpenAI is responding by accelerating its own development of similar technologies, including hiring key talent and focusing on delivering AI agents and coding tools. However, many details about the Spud model remain undisclosed, such as its parameter count, multimodality, or reasoning capabilities. The model’s exact naming—whether GPT-5.5, GPT-6, or something entirely new—is also unknown, but the language used by Altman suggests it could be a breakthrough with AGI-level implications.
In parallel with these developments, there has been a notable breakthrough in AI-assisted mathematics. Terrence Tao, widely regarded as the greatest living mathematician, recently published a paper co-developed with AI models, including an OpenAI model referred to as ChatGPT. Tao used Google’s DeepMind evolutionary AI agent, AlphaEvolve, to help solve complex mathematical problems by splitting proofs between himself and AI. This collaboration highlights AI’s emerging role as a credible co-author and scientific partner, accelerating progress in fields traditionally considered highly challenging for machines.
The video concludes by reflecting on the broader implications of these advancements. Despite skepticism from some quarters, the evidence suggests AI is making rapid, tangible progress in scientific discovery and coding. The speaker encourages viewers to consider whether AI will indeed accelerate the economy and transform industries as predicted by AI leaders. The ongoing developments at OpenAI and breakthroughs like Tao’s collaboration indicate that AI’s trajectory is likely to continue its exponential growth, challenging doubters and reshaping expectations about the future of technology and innovation.
