Google just released Gemini Omni 1.1 Flash, a new version of the Gemini Omni Flash model that first appeared in May and hit public API beta at the end of June.
The new version still generates up to ten seconds of video per run, but for the first time, the Omni API supports video continuation. After a clip is generated, you can now tell the model to keep filming — adding another ten seconds each time, up to a total of 40 seconds per video. Every extension references the previous ten seconds of footage, so characters, motion, and story beats have a decent shot at staying consistent.
Omni 1.1 also adds first-and-last-frame control. Give the model an opening image and a closing image, and it will fill in the continuous video between them. If you want a shot that starts with a person standing in the distance and ends with them right in front of the camera, the model handles the whole movement in between.
There's also a cheaper 360p draft mode. Google says system throughput improves by up to 60 percent, and the cost lands at roughly one-third of 720p. API pricing runs $0.03 per second at 360p, $0.10 at 720p, $0.15 at 1080p, and $0.30 at 4K. The 1080p and 4K tiers are upscaled after generation, not natively produced.