Will Sora Return? OpenAI Plans to Hand Over the Model to Katsberg's Startup, Shifting to a Professional Film Tool Approach

The video generation model Sora, which had been inactive for months, seems to be finding a roundabout way back to the market. According to a report by The Information on October 12, OpenAI is in talks with a stealth startup specializing in AI video technology for film production, planning to hand over the previously discontinued Sora to them, allowing it to reappear in the form of a professional film tool.

The team behind this acquiring company is quite impressive. It was jointly established several months ago by Hollywood producer Jeffrey Katzenberg, former Sora research lead Bill Pibbels, and former Dropbox CFO Soojit Jaswal, and the company's name remains undisclosed to the public. According to the disclosed details of the collaboration, the startup intends to take over and improve the Sora platform, and develop and sell software or services targeting film production companies based on this technology. At present, the specific terms have not been finalized, and whether the company will even obtain the Sora model weights remains uncertain.

Sora's exit wasn't that long ago. Earlier this year, OpenAI shut down the Sora model and the development of a TikTok-like video app to cut costs on non-core projects requiring high computing power, reallocating resources to business and programming tools; its consumer applications and web platform had already been closed, and a cooperation plan worth about $1 billion (approximately RMB 6.7 billion) with Disney was also terminated. In other words, a product once seen as a benchmark for AI video but consuming massive computing power, was voluntarily taken off the table by OpenAI.

OpenAI Wants to Revive Sora: Hand It Over to the Katsberg Team for Professional Film Tools

According to The Information, OpenAI is in talks with a secretive startup specializing in AI video for film, aiming to transfer the previously discontinued Sora, allowing it to be reborn as a professional film tool. This means Sora will shift from a consumer-level toy to an industrial-level tool. OpenAI wants the technology to continue shining without burning money: cutting heavy assets and reducing computing power consumption to focus on enterprise business. This is not the first time Sora has been adjusted.

OpenAI Plans to Restart Sora Through Equity Cooperation, Shifting to the Hollywood Professional Film Production Market

Per The Information, OpenAI is in talks on an equity deal to license Sora tech to a new AI video startup—reviving the shuttered Sora. Founded by ex-Sora research lead Bill Peebles, DreamWorks co-founder Jeffrey Katzenberg, and ex-Dropbox CFO Sujay Jaswa, the plan gives OpenAI equity plus deep model customization beyond standard API. Talks ongoing.....

OpenAI Expands Memory for GPT-6.1 Sol: Default Context Window Up to 500,000 Tokens, Long Texts No Longer Charged Separately

On October 12, OpenAI's Tibo announced that the default context window for GPT-6.1Sol in Codex and Work environments has been increased from 272,000 tokens to 500,000 tokens, a rise of about 80%. He stated that the upper limit will continue to be raised in the future, but this adjustment has already significantly reduced the frequency of "context compression" triggers. This mechanism refers to the model being forced to summarize and discard previous content when its capacity is insufficient, which previously troubled heavy code users.

OpenAI Releases 175-Page Four-Dimensional Proof Manuscript, Solving a Century-Old Mathematical Problem

OpenAI has recently released 722 cutting-edge mathematical manuscripts, including two core papers targeting the century-old mathematical problem of Kakutani's conjecture: one focuses on the three-dimensional Kakutani maximal function conjecture, while the other provides a full-dimensional proof for general four-dimensional Kakutani sets. This breakthrough has pushed the boundaries of multidimensional mathematical understanding and has sparked global academic discussions on the role of AI in advancing fundamental scientific research.

OpenAI Releases a Large Number of Mathematical Research Achievements, Causing a Shock in the Academic Community; Scholars Worry About Disrupting the Tradition of Academic Collaboration

OpenAI released more than 700 mathematical research papers generated by its internal cutting-edge AI model, sparking widespread attention and heated discussions in the mathematics community. Some scholars are excited about the breakthroughs of AI in the field of mathematics, while experts worry about its impact on the academic ecosystem and young researchers. Young researchers' achievements are being affected, and data security is being questioned. Professor Bakmester from New York University criticized OpenAI for releasing such a large amount of research results at once.