Google opens Gemini Omni 1.2 Flash video to businesses, cuts price 30%

The latest:

Video clips of up to 15 seconds can now be generated and re-edited by conversation under Gemini Omni 1.2 Flash, which Google made generally available on November 5, 2026 through the Gemini API, Google Cloud and Gemini for Business. The company said it is the first general release in the Gemini Omni family, and cut the price 30% against the previous version.

Details:

• The pricing: Google set the rate at $0.07 per second of generated video at 720p resolution, a 30% reduction from Gemini Omni 1.1, according to the company. Selected clips can be upscaled afterwards to 1080p or 4K, placing the higher-resolution step after the selection stage rather than at generation.
• Clip length: The maximum clip runs 15 seconds, up from the 10-second ceiling in Omni 1.1, according to Google. Duration is set in one-second increments across a range of 3 to 15 seconds, giving advertisers control over standard ad slot lengths without trimming output after the fact.
• Conversational editing: Google said a generated clip can be modified through multi-turn conversation, including swapping a product, a background, the lighting or the aspect ratio, without regenerating the video from the beginning. That reframes revision as a dialogue step rather than a new production run.
• Brand consistency: According to Google, the model holds faces, clothing, product packaging and brand colors stable across changing camera angles — the failure point that has historically forced commercial users to discard generated footage and reshoot or regenerate it.
• Where it runs: The model is being deployed across Flow, Google Vids and Google Ads, placing it inside the company’s existing advertising and workplace video products rather than in a standalone tool, the company said.
• Consumer rollout: Access through the Gemini app is arriving gradually for Google AI Ultra subscribers, Google said. The company did not give a completion date for that rollout or name the markets covered first.
• Provenance marking: All outputs carry Google’s SynthID invisible watermark, the company said, embedding a machine-detectable provenance signal in generated video as regulators and platforms press for disclosure of synthetic media.
• Acknowledged limits: Google said work continues on improving fast sports motion and complex physics, the two areas it identified as weak in the current model. It did not say when those improvements would ship.

Background:

Gemini Omni is Google’s multimodal model family; version 1.1 preceded this release with a 10-second video ceiling and a higher per-second price. The November 5 launch is the family’s first general availability, moving it out of limited access.

Between the lines:

The combination Google is selling is cost and iteration, not raw quality. A 30% price cut paired with conversational edits that skip regeneration targets the expensive part of commercial video work — the revision cycle — while the company’s own stated weaknesses in fast motion and complex physics point the model toward product shots and static scenes rather than action footage.

What’s next

Watch the pace of the Gemini app rollout to Google AI Ultra subscribers, uptake inside Google Ads among advertisers, and whether Google ships the promised improvements to fast sports motion and complex physics.