OpenAI Is Killing Sora: App Closes April 2026, API Deprecated in September

OpenAI announced on March 24 that it is winding down Sora, the text-to-video AI product that debuted to substantial fanfare in 2024. The shutdown happens in two stages: the consumer application closes on April 26, 2026, and the Sora API — along with all video generation endpoints — is fully deprecated on September 24, 2026. Video generation capabilities are also being removed from ChatGPT as part of the same strategic decision.

Users have been advised to download all content from their Sora libraries before the April cutoff. OpenAI's help documentation confirms that all stored videos and generated images can be exported directly through the web interface.

Why OpenAI Is Walking Away

The stated rationale is resource prioritization. OpenAI is redirecting compute capacity toward products it considers higher-value: enterprise coding tools, its Operator platform for business automation, and a planned unified "super app" that would consolidate multiple OpenAI capabilities under a single interface. Video generation is compute-intensive and, despite the launch attention Sora received, it did not achieve the consumer adoption rates that would justify the ongoing infrastructure cost against those competing priorities.

Analysts framing the decision as evidence of OpenAI's financial maturation are not wrong. The company is two years into pressure to move from research-driven launch cycles toward revenue-disciplined product strategy. Shutting down a capability that doesn't contribute meaningfully to revenue — while it costs significant compute — is exactly the kind of decision a profitable enterprise software company makes. The timing, ahead of an anticipated IPO in 2026, reinforces that reading.

Sora Continues as Research

OpenAI has been careful to distinguish product shutdown from research abandonment. Sora continues as an internal project focused on what the company calls "world models" — AI systems capable of modeling physical dynamics and causality, not just generating plausible-looking video. The stated long-term objective is "automating the physical economy," a framing that positions Sora research within OpenAI's broader AGI-adjacent ambitions rather than the consumer video market.

Whether that distinction holds in practice — and whether world model research retains meaningful funding without a product to drive it — is the question the AI research community will be watching. Competitors including Google (Lumiere, Veo 3), Runway, and Kling continue active development of video generation products, meaning OpenAI's exit creates market space even as it signals internal prioritization shifts.