Google Releases Gemini Omni 1.2 Flash

Google DeepMind announced Gemini Omni 1.2 Flash on November 5, making the video-generation model generally available through the Gemini API and Gemini for business. The company says it creates video clips up to 15 seconds long, supports conversational editing across multiple turns and follows reference images, motion clips and audio cues more closely. Google cut the price 30% from Omni 1.1 to $0.07 per second for 720p video.

Users can choose clip lengths in one-second increments from 3 to 15 seconds, compared with a 10-second maximum for Omni 1.1. Google says teams can draft at 360p or 720p, revise details with follow-up edits rather than regenerate a whole clip, then upscale selected videos to 1080p or 4K. The model can combine character, product or style images, motion clips up to three seconds, and audio references.

Google says Omni 1.2 Flash better preserves characters, objects, scenes, text, clothing and brand colors across frames, camera angles and transitions, reducing visual drift and morphing. Its examples include a 15-second virtual try-on video built from one model photo and front-and-back product shots of three outfits, and an animated poster assembled layer by layer while retaining material, shadow and typography details. The model can also interpolate between opening and closing keyframes or extend a clip into a longer sequence.

Editing can proceed turn by turn, including changing a product, background, lighting, weather or aspect ratio while retaining other elements of the clip. Google also says the model produces more natural dialogue, clearer lip-sync and distinct multi-speaker conditioning, and can create continuous camera movements with fewer unwanted jump cuts. Google notes that work continues on fast sports action, rapid script changes and complex physics.

Google says businesses, startups and creatives are already using the model, which is backed by Google Cloud enterprise infrastructure and security. Provisioned Throughput is available, and generated outputs carry an imperceptible SynthID watermark. The model is rolling out across Google products and developer platforms: developers can use the Gemini API and Google AI Studio; businesses can access it through Google Cloud; and Google says 15-second generation is available in Google Vids at no additional cost with a Google account or Workspace subscription. Fifteen-second video creation in the Gemini app is rolling out to Google AI Ultra subscribers, while Google Ads supports video campaigns. Adobe says Omni Flash 1.2 will bring longer outputs and stronger reference-image consistency to Firefly. Whering’s CEO said tests showed garment drape, color matching and reliability across complex outfits, with lower latency.

Google DeepMind announced Gemini Omni 1.2 Flash on November 5, making the video-generation model generally available through the Gemini API and Gemini for business. The company says it creates video clips up to 15 seconds long, supports conversational editing across multiple turns and follows reference images, motion clips and audio cues more closely. Google cut the price 30% from Omni 1.1 to $0.07 per second for 720p video.

Users can choose clip lengths in one-second increments from 3 to 15 seconds, compared with a 10-second maximum for Omni 1.1. Google says teams can draft at 360p or 720p, revise details with follow-up edits rather than regenerate a whole clip, then upscale selected videos to 1080p or 4K. The model can combine character, product or style images, motion clips up to three seconds, and audio references.

Google says Omni 1.2 Flash better preserves characters, objects, scenes, text, clothing and brand colors across frames, camera angles and transitions, reducing visual drift and morphing. Its examples include a 15-second virtual try-on video built from one model photo and front-and-back product shots of three outfits, and an animated poster assembled layer by layer while retaining material, shadow and typography details. The model can also interpolate between opening and closing keyframes or extend a clip into a longer sequence.

Editing can proceed turn by turn, including changing a product, background, lighting, weather or aspect ratio while retaining other elements of the clip. Google also says the model produces more natural dialogue, clearer lip-sync and distinct multi-speaker conditioning, and can create continuous camera movements with fewer unwanted jump cuts. Google notes that work continues on fast sports action, rapid script changes and complex physics.

Google says businesses, startups and creatives are already using the model, which is backed by Google Cloud enterprise infrastructure and security. Provisioned Throughput is available, and generated outputs carry an imperceptible SynthID watermark. The model is rolling out across Google products and developer platforms: developers can use the Gemini API and Google AI Studio; businesses can access it through Google Cloud; and Google says 15-second generation is available in Google Vids at no additional cost with a Google account or Workspace subscription. Fifteen-second video creation in the Gemini app is rolling out to Google AI Ultra subscribers, while Google Ads supports video campaigns. Adobe says Omni Flash 1.2 will bring longer outputs and stronger reference-image consistency to Firefly. Whering’s CEO said tests showed garment drape, color matching and reliability across complex outfits, with lower latency.