Gemini Omni 1.2 Flash explained: 15 s clips, 30% cheaper

By LLMTest Lab · Nov 8, 2026 · 3 min read videoreleasepricing
On this page

On this page

  1. What is new versus Omni 1.1 Flash
  2. What a clip costs
  3. How it compares on paper
  4. Access and caveats
  5. Who should care

Not tested by us yet. We generate no video in our tests, so every number below comes from Google's launch post or OpenRouter's video model list, and quality claims are Google's own. We'll update this post when independent side-by-side clips against Omni 1.1 Flash and Veo 3.1 appear.

Gemini Omni 1.2 Flash is Google's video generation and editing model with native audio. It extends clips from 10 to 15 seconds and costs about 30% less than Omni 1.1 Flash: $0.07 per second at 720p. If you already pay for Omni 1.1 Flash, the price cut alone is a reason to switch. It arrived one day after 1.1 reached OpenRouter, so 1.1 is now the more expensive and shorter option.

Omni 1.2 Flash Omni 1.1 Flash
Maker Google Google
Released on OpenRouter Nov 5, 2026 Nov 4, 2026
Clip length 3 to 15 s 3 to 10 s
Resolutions 360p, 720p, 1080p, 4K 360p, 720p, 1080p, 4K
720p with audio $0.0695/s $0.1014/s
360p draft with audio $0.0232/s $0.0338/s
4K with audio $0.2085/s $0.3041/s
Inputs Text, image, video Text, image, video

Prices are the per-second rates in OpenRouter's video model list on the day of writing. Google's post states $0.07 per second for 720p and a 30% cut; OpenRouter's numbers work out to 31%.

What is new versus Omni 1.1 Flash

Google lists three changes. Clips run up to 15 seconds in one-second steps. Multi-turn conversational editing now works on full 15-second clips, so you can swap a product or background, change lighting or weather, or switch aspect ratio across turns instead of regenerating. And reference handling is described as better: character, product and style images, motion clips of up to three seconds, and audio cues, with faces, garments and packaging typography said to stay consistent across camera angles.

Google also claims clearer lip-sync, more natural dialogue, multi-speaker conditioning and fewer unwanted jump cuts. There are no benchmark numbers behind those claims in the post, so treat them as vendor-reported.

What a clip costs

Cost per clip is where this release matters. Using OpenRouter's per-second rates:

Clip Price
15 s, 360p draft, Omni 1.2 Flash $0.35
15 s, 720p, Omni 1.2 Flash $1.04
10 s, 720p, Omni 1.2 Flash $0.70
10 s, 720p, Omni 1.1 Flash $1.01
15 s, 4K, Omni 1.2 Flash $3.13

The workflow Google pushes is cheap drafts at 360p or 720p, then an upscale to 1080p or 4K for the clip you keep. At $0.35 for a 15-second draft, ten drafts cost about what one 4K final does.

How it compares on paper

All rates below are per second with audio, from the same OpenRouter list.

Model Rate Max clip
Gemini Omni 1.2 Flash (720p) $0.0695 15 s
Wan 3.0 (720p) $0.10 30 s
Kling v3.0 Pro $0.168 15 s
Veo 3.1 Fast (720p) $0.10 8 s

Omni 1.2 Flash is the cheapest of these at 720p with audio, and it beats Veo 3.1 Fast on length. Wan 3.0 offers twice the maximum duration. Price per second says nothing about quality, and nobody has published a prompt-matched comparison yet. For the open-weights end of the market, see our HyperFlow H3 Flash explainer.

Access and caveats

Omni 1.2 Flash is generally available through the Gemini API, Google AI Studio, Google Cloud (with Provisioned Throughput), Flow, Google Vids, Google Ads and the Gemini app, where it is limited to AI Ultra subscribers. Outputs carry an imperceptible SynthID watermark. Google says fast sports action, rapid script changes and complex physics still need work.

On OpenRouter, video generation only works through the Video Generation API, not chat completions, even though the model also lists text output.

Who should care

  • Teams already on Omni 1.1 Flash. Same resolutions, 30% lower price, longer clips.
  • Ad and social teams that iterate. The 360p draft rate makes trial runs cheap.
  • Anyone comparing per-clip cost. Run your own prompts before trusting the table above, since we have no quality data. Our cost per feature breakdown shows how to turn unit prices into a budget.

Related articles

HyperFlow H3 Flash explained: 13.7B video model, 6 steps
Video Rebirth's HyperFlow H3 Flash shrinks MiniMax-H3 from about 31B to 13.7B parameters, makes video and audio in 4 to 8 steps, and runs on one RTX 4090.
LLM cost per feature in 2026: four workload patterns
Per-feature LLM cost math for 2026: four workload patterns with token budgets, caching discounts, and total USD per 1,000 requests across model tiers.