OpenAI sets April 26, 2026 discontinuation date for Sora video generation product

OpenAI announced the discontinuation of its Sora web and app experiences, effective April 26, 2026, with the Sora API following on September 24, 2026. The company is advising users to export their content before the cutoff, and will permanently delete all associated data after a final export window. Refunds for unused ChatGPT/Sora credits will be honored and can be applied to OpenAI’s Codex product instead.

Why it matters: This is a rare public product cancellation from a frontier lab, and it signals OpenAI’s willingness to kill a high-profile but commercially underperforming generative media offering. Sora launched to massive cultural hype in early 2024 as a text-to-video breakthrough, but never established a clear revenue model, enterprise use case, or distribution moat comparable to ChatGPT, GPT-4o, or even the Codex agent tool. The discontinuation fits a recurring pattern in generative media (segment 09) where early “magic demos” fail to translate into sustainable products due to high inference cost, lack of differentiation, or limited utility. OpenAI appears to be making a capital-compression choice: focus resources on higher-leverage, cash-generating product lines (ChatGPT, Codex, reasoning models) rather than propping up a standalone video tool. The refund language directing credits to Codex is a subtle but telling signal that the company sees agentic coding as a higher strategic priority than consumer video generation.

Grounded expert take: For the AI video generation market, Sora’s shutdown removes one of the segment’s most visible names, likely accelerating consolidation around a smaller set of players like Runway, Pika, and Kuaishou’s Kling. It also raises the stakes for OpenAI’s remaining multimodal efforts (DALL-E, voice, video-in-chat) which will need to demonstrate commercial traction. The larger lesson is one of structural forces: generative media companies face brutal unit economics at scale, and even a front-runner with OpenAI’s brand, research team, and compute resources could not make Sora work as a standalone product. Expect the remaining video AI startups to pitch themselves as integrated into broader agent or enterprise workflows, not as isolated creative tools.