Google Gemini Omni 1.2 Flash adds 15-second AI videos and lower pricing

Google DeepMind is making Gemini Omni 1.2 Flash generally available, according to The Decoder, with longer AI-generated video clips and lower costs. The update matters for creators, marketers and small teams because 15-second clips and iterative editing make short-form video production less fragmented than stitching together many smaller generations.

The company says the model is rolling out across the Gemini API, Google AI Studio, Google Cloud Console, Flow, Google Vids, the Gemini app for Google AI Ultra subscribers and Google Ads. Every generated video includes an imperceptible SynthID watermark, according to Google, to mark its AI origin.

Longer clips, lower 720p costs

Gemini Omni 1.2 Flash can now generate video clips from three to 15 seconds long. The previous maximum was 10 seconds, according to the source. Google also says the updated model costs 30% less than the earlier version, with 720p video priced at $0.07 per second.

For teams producing ads, social posts, product demos or quick concept videos, the extra five seconds are not just a small extension. A 15-second window better matches common creative formats where a scene needs a setup, a product moment and a closing shot. It may also reduce the need to generate several separate clips just to cover one short idea.

The main changes are:

• Clip length increases to a maximum of 15 seconds.
• 720p generation pricing falls 30% to $0.07 per second.
• Generated clips can be adjusted through an ongoing conversation.
• Outputs include an imperceptible SynthID watermark.

For users already working with Google Gemini, the update extends Google’s AI stack further into video production rather than limiting it to chat, search-style help or file work. It also sits alongside Google’s recent push to let Gemini edit generated files inside chat, covered previously by Emonarc in Google Gemini can now edit generated files inside chat.

Conversational editing moves beyond one-shot generation

The most practical addition is what Google calls conversational editing. Instead of starting over when a generated scene is close but not right, users can ask for step-by-step changes to the same clip.

According to Google, those edits can affect elements such as the background, lighting or aspect ratio without regenerating the entire video. The company says this works across the full 15-second clip length.

That workflow could be especially useful for creators who need many small variations: a vertical ad, a brighter version for a product page, or a background change for a different campaign. It also changes how teams may review AI video. Rather than treating each output as a finished take or a failed take, they can work through revisions in a more familiar creative process.

Google describes a tiered production approach intended to reduce costs. Drafts can be created at 360p or 720p, specific details can then be corrected through follow-up editing, and only the final version needs to be upscaled to 1080p or 4K.

Reference inputs aim at brand and character consistency

Google says Gemini Omni 1.2 Flash has improved reference adherence. The model can use character images, product photos, motion clips up to three seconds long and audio cues as inputs, according to the company.

The goal is to keep visible details stable across changing shots. Google says faces, clothing, packaging typography and brand colors are meant to remain consistent as the camera angle changes. For marketers, that is a significant point: AI video is less useful for product or brand work if a logo, garment shape or person’s appearance drifts from shot to shot.

Google’s example is a virtual try-on sequence. From one model photo and product images for three outfits, the model creates a 15-second video with 360-degree turns while preserving the model’s identity and the cut and drape of each garment, according to the announcement.

The company also says overall visual consistency has improved, with less drifting or morphing during movement and camera transitions compared with the previous model.

What creators should watch next

Beyond visuals, Google says the model can produce more fluid, natural-sounding dialogue, clearer lip sync and better separation between multiple speakers. It also says Gemini Omni 1.2 Flash can handle complex continuous camera moves, including FPV drone-style flights and dramatic zoom-outs, with fewer unwanted jump cuts.

For small businesses, the most immediate use cases are likely lightweight ad concepts, product visuals, social video drafts and campaign variations. The availability through Google Ads is notable because it places the model close to a channel where many teams already test creative.

Still, the practical question is how well the model performs in everyday production rather than in selected examples. Creators should test whether reference consistency holds for their own products, whether conversational edits preserve earlier choices, and whether the SynthID watermark fits client or platform requirements.

The bigger pattern is clear: AI video tools are becoming less about a single prompt and more about controlled iteration. Gemini Omni 1.2 Flash’s longer clips and edit loop push Google further in that direction.