OpenAI's Sora dev API and consumer surfaces diverge
OpenAI’s Sora Videos API (endpoints like POST /v1/videos and models sora‑2/sora‑2‑pro) doesn’t fully align with the consumer Sora app/editor and ChatGPT access — a mismatch that’s causing developer confusion about surfaces and capabilities. (blog.laozhang.ai)
OpenAI set firm discontinuation dates: the Sora web and mobile app experiences will end on April 26, 2026, while the Sora Videos API is slated for shutdown on September 24, 2026. (help.openai.com) The Sora consumer app received a built-in editor on March 19, 2026, adding frame‑level trimming, clip stitching, timeline reordering, import-from-drafts, extensions, reprompting segments, and remixing tools. (help.openai.com) Sora’s consumer “Cameo” / character workflow let users create reusable characters from short video+audio captures and choose per-character sharing permissions, functionality documented in OpenAI’s help center. (help.openai.com) OpenAI’s API reference exposed core Videos endpoints (e.g., POST /v1/videos and GET /v1/videos/{id}/download) and the sora‑2 and sora‑2‑pro models, with published per‑second pricing tiers and resolution targets listed on the model pages. (platform.openai.com) Third‑party and mirror documentation that developers rely on flagged feature gaps: some vendor docs and community guides explicitly note that “Cameo” character embedding and certain video‑to‑video/editoric workflows were not available in the public API surface at the time, and API access required org verification or invite workflows. (sora2ai.run) OpenAI published Sora safety guidance on March 23, 2026 and announced the product wind‑down on March 24, 2026, a timeline that coincided with media reports that OpenAI’s planned Disney partnership and related deals were being derailed. (humai.blog)
