An open-source, free client-side playground for the Sora 2 API (www.sora2playground.com)

A new open-source, free client-side playground makes it easy to experiment with OpenAI’s Sora 2 video model from the browser. The app lets users paste their own API key (never stored) and upload an image to guide video generation; everything runs 100% client-side with a privacy-first design—no data collection, no server proxying—so keys and inputs remain local. The project is fully auditable and customizable since the source is public, lowering friction for demos, prototyping, and learning. For the AI/ML community this matters because it removes two common barriers: tooling and trust. Researchers and developers can rapidly try Sora 2’s video capabilities without building backend infrastructure or exposing sensitive data, making it useful for exploratory workflows, reproducible demos, and teaching. Key technical points: the playground invokes Sora 2 via the user’s API key in-browser, supports image-conditioned video generation, and avoids server-side storage or telemetry. Limitations to note: API usage and cost still come from the user’s OpenAI account, and client-side calls depend on browser/network constraints and model latency. Overall it’s a lightweight, privacy-preserving way to test Sora 2 and bootstrap integrations.