OpenAI brings Sora 2 video workflows to developers with reference assets and batch rendering

What OpenAI posted on X

On March 12, 2026, OpenAI Developers said the Video API was gaining new capabilities powered by Sora 2: custom characters and objects, 16:9 and 9:16 exports, clips up to 20 seconds, scene continuation, and batch jobs for generation. The post matters because it framed Sora 2 as a developer workflow, not just a consumer-facing demo. OpenAI was signaling that teams could start treating video generation like a programmable media pipeline.

What the platform docs add

OpenAI’s current video generation guide says the Videos API supports creating new videos from prompts, guiding runs with image references, reusing character assets across generations, extending completed clips, editing existing videos, downloading outputs, and submitting large offline render queues through the Batch API. The same guide says sora-2 and sora-2-pro both support 16- and 20-second generations, while sora-2-pro is the option for 1080p exports in 1920x1080 and 1080x1920.

That set of controls is what moves Sora 2 closer to production use. Reference-guided generation and reusable character assets help maintain visual consistency across multiple shots. Clip extensions and editing reduce the need to regenerate an entire sequence when a team only needs to push a scene a little further or revise one segment.

Why it matters

For product teams, the practical change is throughput. Batch rendering turns Sora 2 from an interactive prompt toy into something that can sit behind a backlog of social clips, ads, explainers, or vertical and horizontal variants. OpenAI’s docs also describe the workflow as asynchronous, which fits render queues and webhook-based pipelines better than a synchronous request-response UX.

There is one strategic caveat. The current docs already mark the Sora 2 video generation models and Videos API for shutdown on September 24, 2026. That means the March announcement still reads as a real expansion of developer control, but it also suggests teams should treat the feature as a near-term production tool rather than a long-horizon platform commitment until OpenAI clarifies the successor path.

Sources: OpenAI Developers on X, OpenAI video generation guide.