Skip to content

Genie 4 reaches the Gemini API, and the pricing is real

Google opens its interactive world model to paid-tier Gemini API developers through the Interactions API, a day after it rolled out to AI Ultra subscribers.

Source: Google

Google made Genie 4 available to developers on 19 November, through the Gemini API’s Interactions API and Google AI Studio. It’s a day after the same model rolled out to Google AI Ultra subscribers via Project Genie.

Genie 4 generates continuous, explorable 720p video at 24fps with synchronised audio from a text prompt, a reference image, or both, and lets you steer the scene in real time with keyboard controls and live text. Access is paid-tier only: no free tier, Batch API, Flex, or Priority inference. Billing runs on emitted video tokens, 240 per frame at 24fps, which Google’s pricing page works out to input at $1.50 per million tokens, text output at $9 per million, and video-with-audio output at $17.50 per million, roughly $0.10 a second. A session keeps billing until you call complete() to end it. Every output carries an imperceptible SynthID watermark.

Why it matters: when Genie 4 launched in October it was a research preview with, in DeepMind’s own words, no pricing, release date or API documentation. That gap is now closed: there’s a real API, a real price, and nothing stopping you building against it today.

The caveat: $0.10 a second adds up fast for anything exploratory, and an open session bills continuously whether you’re actively steering it or not.

Share

Related reading

More from The Wire

All briefs

A Linux KVM bug let one VM read 127GB of host memory

A bug bounty researcher used a four-year-old stale pointer in nested virtualization to read host kernel memory from inside an ordinary cloud VM. DigitalOcean disclosed and patched it as CVE-2026-99312.

via DigitalOcean