AI News · 4 min read
Gemini Omni 1.2 Flash: 15-second video, 30% cheaper
Google DeepMind shipped Gemini Omni 1.2 Flash on Nov 5, 2026 — native 15s clips, multi-turn edits, $0.07/s at 720p. What’s new and who should care.
What Google shipped on November 5, 2026
Google DeepMind published Introducing Gemini Omni 1.2 Flash ↗ on November 5, 2026, with matching notes on the Google Cloud blog ↗ and a model card dated 5 November 2026 ↗. The Verge’s Emma Roth ↗ covered the same day.
The short version of what GA’d on the Gemini API and Gemini for business / Google Cloud:
Native video up to 15 seconds (was 10s on Omni 1.1). Duration is selectable from 3–15 seconds in 1-second increments.
Price 30% lower vs Omni 1.1: $0.07 per second for 720p, per Google’s launch post.
Multi-turn conversational editing without a full re-generation of the clip.
Better reference adherence for images, short motion clips (≤3s), and audio cues.
Draft at 360p / 720p, then upscale to 1080p or 4K.
More natural speech / lip-sync / multi-speaker delivery, plus continuous camera moves with fewer jump cuts.
SynthID watermarking; Provisioned Throughput available for production traffic.
Product surfaces called out by Google:
Google Vids: 15s clips at no additional cost with a Google account / Workspace.
Gemini app rolling out to Google AI Ultra.
Also landing in Flow and Google Ads.
Google is explicit about remaining limits: fast sports, rapid script changes, and complex physics are still improving. Treat that as the vendor’s own caveat, not our lab result.
What people are saying on X
Official framing from Google AI:
“Meet Gemini Omni 1.2 Flash… bumped the video generation and editing limit up to 15 seconds… dropped the price by 30%… $0.07 per second for 720p video.”
— @GoogleAI ↗, Nov 5, 2026 (~1.8k likes / ~173k impressions at fetch)
Arena’s public leaderboard claim (their ranking, not our independent bake-off):
“Gemini Omni 1.2 Flash… debuts at #1 in both Text-to-Video and Image-to-Video, and #3 in Video Edit! … #1 in Text-to-Video with 1541 pts. This is a +27 pt improvement from the previous Gemini Omni 1.1 Flash at #2 with 1514 pts.”
— @arena ↗ (Arena.ai)
Skeptical creator bake-off — duration and conditioning caveats:
“Still not there yet… Omni gave me 10 seconds, Seedance 15… Google says Omni now goes up to 15 seconds, but only from text or a single starting image. Add a last frame or reference images and its back to 10… Seedance 2.5… goes up to 30 seconds, and its still ahead.”
Availability note from DesignArena:
Now listed on DesignArena with multi-turn editing and upscale to 1080p / 4K.
We did not find a strong Omni 1.2–specific Reddit thread at research time, so we are not inventing Reddit quotes. X + official posts + The Verge are the ground truth for this desk pass.
Our take: who should care
API / product teams already on Omni 1.1: the 15s ceiling and 30% price cut are the practical unlock. If your ads, product demos, or storyboard loops were stuck at 10s, re-price a 12–15s brief at $0.07/s (720p) before you rewrite every prompt. Multi-turn edit without full re-gen matters more than the scoreboard for iteration cost.
Creators in Gemini / Vids / Flow: Google Vids’ “no additional cost” 15s clips (with a Google account / Workspace) are the lowest-friction path. Ultra users should watch the Gemini app rollout; do not assume every surface is live on day one.
Compared with rivals (without inventing our own benches): Arena has Omni 1.2 Flash at #1 Text-to-Video / Image-to-Video on its published board — attribute that to Arena. Creator reports (see @aipulseda1ly) still flag conditioning limits (last-frame / multi-reference dropping back toward 10s) and longer ceilings elsewhere (e.g. Seedance 2.5 up to 30s in that post). Run your own brief: text-only 15s, single start-frame 15s, then last-frame + references, and log what duration you actually get.
Still improving, per Google: sports, rapid script changes, complex physics. Do not plan launch creative that depends on those until you have signed-off samples.
For the wider video-generator landscape, see our best AI video generators 2026. For where Gemini sits among assistants, see ChatGPT vs Claude vs Gemini. Adjacent still-image workflow notes live in our ChatGPT image generation 2026 guide.
AdSense-safe note: we are documenting what Google shipped and how to evaluate it — not recommending any paid Google plan upsell.
Sources
Google: Introducing Gemini Omni 1.2 Flash ↗ — Nov 5, 2026
DeepMind model card: Gemini Omni 1.2 Flash ↗ — Published 5 November 2026
The Verge — Emma Roth ↗ — Nov 5, 2026