ToolsAugust 25, 2026via Vercel Blog
AI Gateway now supports asynchronous video generation
Why it matters
A critical infrastructure feature for production AI apps: asynchronous video generation eliminates timeout failures and enables better scaling patterns for long-running workloads. Practitioners building video features into products need this.
Key signals
- Vercel AI Gateway adds async video generation support
- Four options: webhook delivery, polling, start-and-retrieve, or streaming (generateVideo default behavior preserved)
- Solves timeout problem for video jobs taking seconds to minutes
- Works with text-to-video, image-to-video, reference-to-video inputs
- Operations are JSON-serializable for database/queue storage
- Monitoring via AI Gateway Logs with async-only filter
- Requires latest AI SDK and AI Gateway provider versions
- URL or inline bytes returned; hosted URLs can expire
- Vercel AI Gateway adds asynchronous video generation support
- Four workflow options: webhook, polling, deferred retrieval (startVideo/getVideoStatus), and standard generateVideo (unchanged)
- Addresses multi-second/multi-minute video generation timeouts in synchronous HTTP requests
- Async jobs appear in AI Gateway Logs; operations are JSON-serializable for database/queue storage
- SDK auto-downloads provider-hosted videos for webhook/polling workflows; getVideoStatus returns provider URLs (which can expire)
- Feature requires latest AI SDK and AI Gateway provider versions
The hook
Vercel's AI Gateway now lets video generation run async—no more hung requests. Webhook, poll, or start-and-retrieve later.
on AI Gateway can now run asynchronously. Video generation
By default, keeps one HTTP request to AI Gateway open until the result is ready. Because video generation can take seconds or minutes, that request can exceed request timeouts.generateVideo
With asynchronous generation, your application can…