Sora 2

OpenAI

Sora 2 (sora-2) is an OpenAI model that generated video with synchronized audio, released on September 30, 2025. OpenAI discontinued the Sora web and iOS apps on April 26, 2026, and sora-2 was available via the Videos API until it shut down on September 24, 2026. OpenAI's deprecations page lists no replacement.

Full Specifications

All specsBest valueCheapestHeavy workloadMax capabilityDifferences only

	Sora 2OpenAI Retired - app discontinued April 26, 2026; API shut down September 24, 2026
Identity
Developer	OpenAI
Released	Sep 2025
Retired	24 September 2026
Status	Retired. OpenAI's deprecations page lists sora-2 and the Videos API as shut down on 24 September 2026, with no replacement listed.
Licence	Proprietary
Self-hostable	No
Cost
Price per second	Discontinued - the Sora app ended April 26, 2026 and the API shut down September 24, 2026
Cheapest plan	Not offered - no consumer subscription plan exists since the Sora app was discontinued; API access was metered pay-per-second, not a subscription tier, and is itself winding down
Free tier	Not offered - no free tier is currently available; the Sora app (which had ChatGPT-plan-linked free generation limits while live) was discontinued April 26, 2026
Capacity
Max duration	20 seconds per single generation (selectable durations of 4, 8, 12, 16, or 20s) - "Both sora-2 and sora-2-pro support 16- and 20-second generations"; extendable up to 6 times via the separate video-extensions endpoint, for a maximum stitched length of 120 seconds
Max resolution	720p - landscape 1280x720 or portrait 720x1280 (the size/pricing the model's own page lists for base "Sora 2"); 1080p (1920x1080 or 1080x1920) requires the separate, higher-priced sora-2-pro model
Capability
Text-to-video	Yes - primary interface is a natural-language text prompt (prompt in, video+audio out)
Image-to-video	Yes - an input_reference image (image/jpeg, image/png, or image/webp, matching the target video's resolution) can be supplied as the video's first frame via the Videos API
Native audio	Yes - synchronized dialogue, sound effects and ambient audio are generated natively alongside video in a single call (Audio is listed as an Output-only modality on the model's own page); Azure's listing adds that audio is available in multiple languages
Camera controls	Prompt-based only - no dedicated camera-control API parameter; camera direction (pans, tracking shots, angles) is specified via natural-language prompt text, per Azure AI Foundry's Sora 2 listing ("detailed prompt understanding for studio shots, scene details, and camera angles")
Rights
Commercial rights	Yes, for content generated while the product was live - OpenAI's Terms of Use assign full ownership of Output to the user ("you ... own the Output ... We hereby assign to you all our right, title, and interest"), subject to Sora-specific Usage Policies restrictions (no copyrighted characters/music, no depictions of real people/public figures); moot for new generations now that the consumer app is discontinued and the API is winding down
Access
API	Retired - the Videos API (POST/GET /v1/videos and related endpoints) shut down on September 24, 2026; OpenAI lists no replacement for sora-2, sora-2-pro, or their dated snapshots
Chat app	Discontinued - the standalone iOS "Sora" app and the sora.chatgpt.com web experience were both discontinued April 26, 2026
Cloud marketplaces	Microsoft Azure AI Foundry - launched as a public preview October 15, 2025 at the same $0.10/sec (720p) pricing as the OpenAI API; not found on AWS Bedrock or Google Cloud Vertex AI; tied to the same underlying OpenAI model, so not confirmed to survive the API's September 24, 2026 shutdown
Fine-tuning	Not offered - no fine-tuning endpoint or workflow is documented anywhere in the Videos API guide or reference for sora-2/sora-2-pro (fine-tuning support on OpenAI's platform is limited to text/image models)
Measured quality
LMArena Elo	1343 (checked Nov 2026)
LMArena rank	Rank 17