Sora AI動画生成

Sora is OpenAI's text-to-video diffusion model, first announced in 2024 and shipped in production as Sora 2 — and the higher-fidelity Sora 2 Pro — in 2026. It generates 4–12 second clips from a single text prompt, with native synchronized audio. Dialogue, ambient sound, music and on-screen text are part of the same generation rather than added in post.

The two variants share the same underlying model family but trade off speed and price for fidelity. Sora 2 is the everyday workhorse: 720p output at 3 credits per second, fast turnaround, suitable for iteration. Sora 2 Pro adds 1080p output, longer context for complex scenes, and is the choice for hero shots and final deliverables at 14 credits per second.

Both variants support image-to-video — you supply a first frame and the model animates from it. This is useful for product photography, character continuity and storyboarded shots where the look of frame one is non-negotiable.

Typical use cases: short-form social content (TikTok / Reels), product hero videos, music-video B-roll, advertising mood boards and concept teasers. Sora is particularly strong at human faces, physically plausible motion and indoor lighting; it is weaker than Seedance for highly stylized animation and weaker than VEO 3.1 Quality for ultra-fine textural detail.

On FlyAIgh, Sora 2 and Sora 2 Pro both ship under a single credit-based wallet — no separate OpenAI subscription, no API keys to manage. A 5-second Sora 2 clip costs 15 credits; a 5-second Sora 2 Pro clip costs 70 credits.