AI

Gemini Omni 1.2 Flash: 15-second clips and 30% cheaper video

Key takeaways

  • Clips now run to 15 seconds natively, up from 10 in Omni 1.1
  • Price falls 30% to $0.07 per second of 720p, about $1.05 a clip
  • Multi-turn editing works across full clips via previous_interaction_id
  • No benchmark scores for 1.2 are published, so test it on your own briefs

Google has released Gemini Omni 1.2 Flash, a video generation and editing model that produces clips of up to 15 seconds natively and costs 30% less than Omni 1.1 Flash. It is generally available on the Gemini API and Google Cloud, with fewer re-renders and a lower cost per second.

The model comes from Google DeepMind. It accepts text, images, short video clips and audio as input and outputs video with audio. Editing is handled through the Interactions API, which Google's developer docs describe.

What it does well

The headline change is length. Generation and editing now reach 15 seconds, up from 10 in Omni 1.1, and duration can be set in 1-second steps from 3 to 15 seconds. For ad work that matters: QuickFrame AI says 15 seconds at 1080p lets users build a full connected TV (CTV) ad in one generation.

Google lists six upgrades in total. Beyond length and duration control, they are:

  • Reference adherence: character, product or style images can be combined with motion clips of up to 3 seconds.
  • Consistency: faces, garments, packaging text and brand colours are meant to hold across camera angles with less morphing.
  • Speech: dialogue sounds more natural, with clearer lip-sync and distinct multi-speaker conditioning.
  • Camera control: FPV flythroughs, zoom-outs and orbits show fewer unwanted jump cuts.

The most practical feature is multi-turn conversational editing, which now works across full 15-second clips. A team can swap a product, background, lighting or aspect ratio turn by turn. The docs show this through a previous_interaction_id, so the earlier video is not re-uploaded.

Google frames cost control as a workflow: draft at 360p or 720p, fix details with follow-up edits, then upscale the best result to 1080p or 4K.

Where it falls short

Google itself says fast sports action, rapid script changes and complex physics still need work. Those are not edge cases for anyone making sports promos or fast-cut social content.

Evidence is thin. Google has published no benchmark scores for version 1.2, and its claims rest on side-by-side demos against Omni 1.1. The DeepMind model page reports human-rater results for the Omni Flash line, including 1,003 MovieGenBench prompts, and an earlier Omni Flash version took first place on the Arena text-to-video board in June. Those results predate 1.2. Partner praise, such as invideo's VP of Engineering saying the model sticks to the prompt and that physics and sound feel real, is self-selected.

It is also Google-hosted only. There are no open weights and no local deployment, and audio references remain allowlist-gated on Google Cloud.

How it compares

The source gives pricing and specifications only for Google's own models, so the useful comparison is against its predecessor.

FeatureOmni 1.1 FlashOmni 1.2 Flash
Maximum clip length10 seconds15 seconds
Price per second, 720pAbout $0.10$0.07
Duration controlNot stated3 to 15 seconds, 1-second steps
Editing across full clipUp to 10 secondsFull 15 seconds, multi-turn
Published benchmarksFamily results onlyNone for 1.2

Rivals' video models are not covered in the launch material, so buyers should run their own briefs through competing services before committing.

Pricing and value

Google charges $0.07 per second of 720p video, and Google says the savings are similar at other resolutions. A full 15-second 720p clip costs about $1.05.

The value argument rests on the draft, edit, upscale workflow rather than the sticker price alone. Drafting cheaply and fixing one detail by edit means fewer full re-renders, which Google presents as the way to control cost. Provisioned Throughput is available, and Google Vids users can generate 15-second clips at no additional cost.

For IT teams, the enterprise route is the Agent Platform API on Google Cloud, which comes with enterprise SLAs. Prototyping can start in Google AI Studio. The model also surfaces in Flow, Google Vids, Google Ads and the Gemini app, which is rolling out to AI Ultra. Our look at how cost per task diverges from headline pricing applies here too: per-second rates only tell part of the story once re-renders are counted.

Who should use it

Good fits include ad agencies and in-house marketing teams producing short spots, e-commerce teams that need product swaps and background changes on the same clip, and developers building video features into apps through the Gemini API. The 15-second ceiling suits CTV and social formats directly.

Teams that should wait or skip: sports and fast-action producers, anyone needing complex physics, and organisations that require on-premises deployment. Anyone choosing on benchmark evidence will find none for 1.2.

Verdict

Worth a trial for teams already producing short-form video at volume. The lower price and full-length editing cut real waste, and the Google Cloud route brings enterprise SLAs. The missing 1.2 benchmarks and Google's own admission on fast action mean the sensible move is a small paid pilot on your actual briefs, comparing re-render counts against Omni 1.1 Flash before committing budget.

Frequently asked questions

How much does Gemini Omni 1.2 Flash cost?

Google charges $0.07 per second of 720p video, about $1.05 for a full 15-second clip. That is roughly 30% below Omni 1.1 Flash, which bills at around $0.10 per second.

Can Gemini Omni 1.2 Flash run locally?

No. It is hosted by Google only, through the Gemini API and Google Cloud. There are no open weights and no local deployment.

More from Future Technology HQ