Google launches Gemini Omni 1.2 Flash with cheaper, longer video

Google launched Gemini Omni 1.2 Flash on Nov 5, making the video model generally available across the Gemini API and its business offerings. The release increases native clip length to 15 seconds, adds multi-turn conversational editing and cuts the listed price for 720p generation to $0.07 per second. Google says that is 30% lower than Omni 1.1.

The change brings a longer usable shot into the same generation workflow. Google says developers can request durations in one-second increments from three to 15 seconds, compared with a 10-second maximum for Omni 1.1. For teams making explainers or product demonstrations, that expands how much action can fit into one clip without stitching separate outputs together.

The price needs its unit attached. At the announced 720p rate, a 15-second generation works out to $1.05, our multiplication of Google's per-second figure. That is not a quote for a finished production: retries, additional edits and any separately charged processing can change the bill. Google describes drafting at 360p or 720p and upscaling selected results to 1080p or 4K, but the announcement does not establish that every stage costs the same.

Editing is the other major workflow change. Rather than starting from a fresh prompt each time, users can refine a clip through follow-up instructions, including changing its background, lighting, weather or aspect ratio. Google says the conversational editing feature works across full 15-second clips. The practical question is whether an edit preserves the parts a creator wanted to keep, not simply whether the model accepts another instruction.

Google also claims stronger reference following and consistency across camera angles, with less visual drift. Inputs can combine character, product or style images, short motion clips and audio cues. These are the company's descriptions and selected examples, not an independent reliability benchmark. Google acknowledges continuing work on fast sports action, rapid script changes and complex physics.

Availability differs by product. The announcement lists the Gemini API and Google Cloud for developers and businesses, with Provisioned Throughput available. It says outputs carry an imperceptible SynthID watermark. Consumer distribution includes Flow and Google Vids, while 15-second generation in the Gemini app is rolling out to Google AI Ultra subscribers; general availability of the model should not be read as identical access in every app or subscription.

Builders already connecting generation tools to coding agents can compare the release with the existing Gemini API CLI workflow. As with other model API pricing, the useful comparison is the cost of an acceptable result, including failed attempts, rather than the lowest advertised unit price alone. Our AI models explained guide is the place to read Gemini's other models, including Carbon and Argon, against ChatGPT and Claude.