Google has launched preview webhook support for the Gemini API, giving developers a push-based way to track long-running jobs instead of repeatedly polling for status. According to Google's May 4 changelog entry and product post, the feature is meant for asynchronous workflows where model work may take minutes or longer.

What changed

Until now, developers commonly had to keep calling Gemini operation endpoints to see whether a job had finished. Google says the new webhook flow sends an HTTP POST to a developer-controlled endpoint as soon as a supported task completes, cutting down on polling overhead and reducing delay between completion and downstream processing.

The documentation says webhook support is available for Batch jobs, Interactions, and video generation workflows. Google also supports two setup modes: static webhooks configured at the project level and dynamic webhooks attached to individual requests when teams want different jobs routed to different endpoints.

Why it matters

For teams building agents, background pipelines, or high-volume content workflows, this is a practical infrastructure upgrade rather than a model launch. Google says webhook requests are signed using the Standard Webhooks spec, with headers such as webhook-signature, webhook-id, and webhook-timestamp. The company also says failed deliveries are retried for up to 24 hours.

The conservative takeaway is simple: Gemini is becoming easier to integrate into systems that already depend on event-driven backends. That should matter most to developers running batch processing, long video jobs, or chained agent workflows where polling adds cost and operational noise.