Sora

Editor's Review

Sora is dead — this is an obituary, not a review. The timeline: March 24, 2026, OpenAI announced the Sora shutdown (the planned $1B Disney partnership died with it); April 26, the consumer web and mobile app was discontinued; September 24, the Sora 2 / Sora 2 Pro models and Videos API were permanently removed — "no one-to-one replacement"; October 15, Microsoft Foundry retired its last Sora deployment. The WSJ's postmortem: infrastructure cost ~$1M/day against $2.1M lifetime revenue, roughly $1.30 in compute per 10-second clip. Altman redirected compute to enterprise tools, coding, and AGI, calling Sora a "side quest." While it lived, Sora 2 (launched September 2025) had added synchronized audio — its one genuine edge — but it never launched in promised markets like India, capped generations at 20 seconds, heavily blocked real people and copyrighted content, and offered "generate and hope" with limited editing control. Historical pricing for migration math: Sora 2 $0.10/second at 720p; Sora 2 Pro $0.30 (720p) / $0.50 (1024p) / $0.70 (1080p); remaining user credits were moved to Codex use. Why keep the page? Migration signpost. Your honest alternatives: Runway (the maintained all-in-one platform, $15/month Standard), Google Veo 3.1 (native audio + 4K, the quality pick), Kling 3.0 (the budget pick, native lip-sync), Luma Ray3 (atmosphere). Sora proved text-to-video could be magical; it also proved magic without unit economics is just a demo with a datacenter bill.