OpenAI Spud AI Model Takes Shape as Sora Exits and Focus Narrows

A conceptual image within a modern server room, where a holographic video film strip of a woman's face is connected to a central sphere containing the cyan OpenAI geometric logo. Yellow fiber-optic tubes extend from the sphere to a complex, cubic data structure, illustrating the transformation of video into actionable data.
An interpretation of AI processing, showing a central OpenAI logo sphere connecting a video film strip with a futuristic data lattice in a server room.

OpenAI is preparing a new AI model codenamed Spud, according to reporting by Stephanie Palazzolo and Amir Efrati at The Information, as the company simultaneously shuts down its Sora video app and tightens its product focus. The decisions arrive at a moment when compute constraints and competitive pressure are forcing the company to make hard choices about what to build — and what to drop.

What the OpenAI Spud AI Model Means for the Company’s Direction

Details on Spud remain sparse. OpenAI has not disclosed the model’s capabilities, intended use cases, or release timeline. What is known is that the Spud model is being developed alongside a broader internal reorganization, with CEO of Applications Fidji Simo taking on a more central operational role.

Simo has been explicit about the stakes. At a recent all-hands meeting, she told staff: “We cannot miss this moment because we are distracted by side quests.” One of the first casualties of that philosophy is Sora, OpenAI’s AI video generation app, which the company announced it will discontinue.

The restructuring also includes plans for a desktop “superapp.” OpenAI is planning a desktop ‘superapp’ that merges ChatGPT, the Codex AI coding tool, and the Atlas browser into a single interface, according to The Verge. What the app might include beyond those three products remains an open question.

Sora’s Exit: Six Months, One Billion Dollars, and a Quiet Retreat

CNET reports that Sora was available for just 176 days — roughly six months — before OpenAI pulled the plug. That short lifespan belies the scale of the partnerships it leaves behind. Creative Bloq notes the shutdown cancels a $1 billion deal to bring user-prompted AI-generated content to Disney+.

When Sora 2 launched in September 2025 — adding audio capabilities and more accurate physics — it was described in some corners as “the end of Hollywood.” That framing now looks like overreach. Usage data from Higgsfield, a video generation platform for creators and agencies, shows Sora peaking around its October 2025 launch before declining steadily, according to Taz Patel, Higgsfield’s VP of platform partnerships.

The ad industry, for one, had already moved on before the official shutdown. “The truth is we had already moved on to other tools that better fit the way our creative teams work,” said Tim McCraken, SVP creative and AI at BarkleyOKRP, speaking to ADWEEK. The departure of Sora will also remove it as a partner model in platforms like Adobe Firefly.

Content moderation played a role in the product’s troubled arc. OpenAI CEO Sam Altman claimed surprise that users “don’t want their cameo to say offensive things or things that they find deeply problematic” — a comment that came as the company shifted to an opt-in policy for protected intellectual property.

The Competitive Picture: Anthropic Is Winning Enterprise Buyers

The backdrop to all of this is a deteriorating position in the enterprise market. Data drawn from more than 50,000 customers of payments startup Ramp shows that first-time enterprise buyers are now choosing Anthropic’s services three times as often as OpenAI’s — a sharp reversal from the year prior, according to TechSpot.

Anthropic, founded in 2021 by former OpenAI researchers, has taken a more deliberate approach to enterprise positioning. That focus appears to be paying off precisely as OpenAI is making its own push in that direction. Bernard Golden, CEO of Silicon Valley consultancy Navica, has noted that surging compute demand is forcing OpenAI to funnel resources toward its greatest revenue driver — ChatGPT.

The AI video space itself hasn’t stood still. Meta launched its own AI video app, Vibes, shortly before OpenAI released Sora — a reminder that OpenAI was never alone in the consumer video market. With Sora gone and Google and Apple also active in adjacent AI product areas, OpenAI’s consumer experiments are narrowing while its enterprise ambitions expand.

What Remains Unknown About Spud and What Comes Next

The Spud model is the most significant unknown in OpenAI’s near-term roadmap. No public information exists yet about its architecture, target users, or how it fits into the superapp strategy alongside ChatGPT, Codex, and Atlas. Whether it will be positioned for consumer or enterprise use — or both — has not been disclosed.

Other open questions are equally material. What additional products might OpenAI classify as “side quests” and wind down? How will the loss of the Disney+ deal affect OpenAI’s media partnerships more broadly? And can the company close the enterprise gap with Anthropic while simultaneously executing a multi-product consolidation?

For enterprise buyers and developers currently evaluating AI vendors, the message from OpenAI’s current posture is one of concentration rather than breadth. Whether Spud accelerates that story or complicates it will depend on details the company has yet to share.

FAQ – Frequently Asked Questions

How will OpenAI’s new desktop ‘superapp’ integrate with existing productivity tools?

The desktop ‘superapp’ is expected to support integrations with popular productivity software such as Microsoft Office and Google Workspace, allowing users to access OpenAI’s features directly within their existing workflows. This integration is likely to be achieved through APIs and software development kits (SDKs) that will be made available to developers. The exact details of these integrations are still unclear, but they are expected to be announced in the coming months.

What specific enterprise needs is Anthropic’s AI model addressing that OpenAI’s is not?

Anthropic’s AI model has been designed with a focus on explainability and transparency, which is particularly important for enterprise customers who require a high level of accountability and interpretability in their AI solutions. Additionally, Anthropic’s model has been tailored to meet the specific needs of industries such as finance and healthcare, where data privacy and security are paramount. This targeted approach has helped Anthropic gain traction with enterprise buyers.

Will OpenAI continue to support developers who have built applications on top of Sora?

OpenAI has announced plans to provide support for developers who have built applications on top of Sora, including access to alternative AI models and tools that can help them transition their applications to new platforms. The company is also expected to offer guidance and resources to help developers navigate the changes and minimize disruption to their businesses. A dedicated support portal is likely to be launched in the coming weeks.

Laszlo Szabo / NowadAIs

Laszlo Szabo is an AI technology analyst with 6+ years covering artificial intelligence developments. Specializing in large language models, ML benchmarking, and Artificial Intelligence industry analysis

Categories

Follow us on Facebook!

Banner showing “Put Claude to work on your computer” representing Anthropic Claude’s computer use feature on Mac systems.
Previous Story

Anthropic Claude Computer Use Arrives on Mac—But Who Gets Left Out?

A conceptual illustration showing digital twins in healthcare being hindered by a significant "data readiness gap." On the left, a chaotic assortment of fragmented data types, including physical medical charts, scattered papers, and broken lab icons, attempts to flow across a deep chasm towards an incomplete digital model of a human. This idealized digital twin on the right is only partially formed within a futuristic, crumbling grid framework, while labels point to specific missing components and integration failures, visually representing the challenges faced in creating fully functional digital twins in healthcare.
Next Story

Digital Twins in Healthcare Face a Data Readiness Gap

Latest from Blog

Go toTop