Inside the rise and fall of OpenAI's Sora, which worked apart from OpenAI's research team, as the company shuts down Sora and redirects compute to other tasks
When Sam Altman arrived in Los Angeles to attend Vanity Fair's Oscar afterparty earlier this month, his company was just weeks away …
The closure follows OpenAI's announcement that it would end the consumer app, developer offering, and ChatGPT video feature built on Sora. Reports that some team members were surprised by the decision underline the abruptness of the wind-down.
First-order effects
Users and developers of Sora-based products lose those OpenAI video tools as the company discontinues them.
OpenAI reallocates compute previously assigned to Sora, and the separately run Sora effort is no longer a standalone internal priority.
Second-order effects
Developers and creative users that incorporated Sora must shift video-generation workflows to other tools or remove the feature, creating a near-term migration burden.
The move concentrates OpenAI's scarce compute and product attention on its remaining tasks, rather than maintaining a dedicated video product line.
Third-order effects
If frontier labs continue pruning expensive, standalone modalities, compute allocation may become a more decisive product-selection mechanism than a model's visibility or early cultural impact.
Sora's path—from disputes over creative-professional treatment and training-data opacity to product closure—also shows how deployment governance can shape whether generative-media products remain viable.
The trend: Frontier AI labs are increasingly concentrating compute and organizational capacity on a narrower set of strategically prioritized products.
The FT is reporting that OpenAI's ‘adult mode’ Chat has also been dropped as a side-quest along with Sora, and has been shelved indefinitely. This was very close to release. They appear to be focusing everything on the new model, which will arrive in about two weeks.
@WSJ I'm glad those random diversions are gone. Hopefully, OpenAI will now focus on becoming the smartest AI. They should have followed Anthropic's singular focus on building a highly useful LLM instead of diverting into side projects that most high paying users don't care about.
The fact that Anthropic (reducing limits) and OpenAI (killing Sora, redirecting GPUs) can't keep up with usage despite Gemini's growth suggests there's sufficient demand to support at least three big AI labs and at worst, if one lab faltered, its infrastructure assets and
Workers should never forget how quickly and easily the managerial class fell for AI solely because they thought it would replace the workers right away.
This implies a fine tuned dirty-chatty trained on all the erotica on the internet, code name ‘Slaanesh.’ Only accessible through a single terminal in a hypergoon chamber deep below OpenAI HQ. Very few have clearance.
OpenAI's new AI model coming soon. Bullish for imminent AI model advances as Anthropic's Mythos blog post also leaked last week WSJ: “OpenAI was weeks away from finishing work on a new AI model, code-named Spud, and needed to free up more computing resources to power the coding
i think being ‘just online enough’ is a major skill — for journalists, politicians, companies. for politicians, you don't want to be offline (the vibes matter) but you don't want to mistake viral moments for the pulse of popular opinion same for firms. i think maybe openAI was [i…