Gemini Omni 1.2 Flash goes GA: 15-second clips, multi-turn edits, $0.07/s at 720p

Google has put Gemini Omni 1.2 Flash into general availability. The DeepMind twin URL resolves to the same Google Blog post. Google says this is the first time Gemini Omni is generally available, with enterprise-grade infrastructure on Google Cloud, Provisioned Throughput, and SynthID watermarks on outputs.

This is a product update on the same line we compared with Utopai X in AI movie models: Gemini Omni vs Utopai X, not a new series episode. For open-weight context from the same week, see MiniMax H3 vs HunyuanVideo 1.5.

What 1.2 Flash adds

• Longer clips. Native generation and editing up to 15 seconds, up from a 10-second max on Omni 1.1. Duration can be set in one-second steps from 3 to 15 seconds.
• Multi-turn editing. Follow-up prompts can swap a product or background, change lighting or weather, or switch aspect ratio on the same clip, including full 15-second clips, instead of regenerating from scratch.
• Reference conditioning. Character, product or style images, short motion clips of up to three seconds, and audio cues can guide a generation. Google says faces, garments, packaging typography and brand colours stay more consistent across angles.
• Multi-speaker lip-sync. Google claims clearer lip-sync and distinct multi-speaker conditioning for dialogue scenes in Flow and Google Vids.
• Draft then upscale. Draft at 360p or 720p, refine with follow-ups, then upscale the keepers to 1080p or 4K.

Price and surfaces

Google prices 720p at $0.07 per second, which it says is 30% below Omni 1.1. We are not quoting other resolutions here until a pricing page line is checked for them.

Surfaces named in the post: the Gemini API, Google Cloud (including Provisioned Throughput), Flow for directing and multi-turn edits, Google Vids (15-second clips at no additional cost with a Google account or Workspace subscription), the Gemini app rolling out to Google AI Ultra subscribers, and Google Ads.

Limits Google admits

Google says work continues on fast sports action, rapid script changes and complex physics. Treat those as weak spots until proven otherwise.

Why this matters

Fifteen-second native clips and multi-turn edits matter more for ads and product explainers than another arena score. The draft-at-cheap-resolution path is the cost control: fix a detail with a follow-up rather than paying for a full re-roll, then upscale. Pair it with the H3 vs HunyuanVideo open-weight options when you need local or licence-constrained video, and with disposable agent runtimes such as Render Sandboxes when the pipeline is code, not a clip.