Google DeepMind launches Gemini Omni 1.2 Flash GA: 15-second video, multi-turn edit, 30% cheaper

Why it matters

【Key points】

• Google/DeepMind’s 2026-11-05 17:00Z primary launches Gemini Omni 1.2 Flash as GA on the Gemini API and Google Cloud. vs Omni 1.1: native clips up to 15s (3–15s in 1s steps)
• multi-turn conversational editing
• stronger reference adherence (images, ≤3s motion, audio)
• ~30% price cut (~$0.07/s at 720p)
• draft at 360p/720p then upscale to 1080p/4K.
• Changelog same day; 1.1 Flash auto-routes from 2026-12-03.
• SynthID + Provisioned Throughput.
• Rolling into Flow, Vids, Gemini app (Ultra), Ads.

【Our read】

Background

• Omni is Google’s video gen/edit line (not Genie world-sim).
• 1.2 Flash packages longer shots, conversational revisions, consistency, and a price cut for production ads/try-on/social workflows rather than a pure leaderboard claim
• physics/fast action still improving.

Peers

• Images 2.6 (e/1088) just productized GIF/edit for stills; Kling/Runway/Luma/Sora keep pressure on duration/consistency.
• Google answers with 15s + multi-turn edit + Cloud GA and a Dec 3 migration window.
• No full external video Arena numbers on the primary.

Impact

immediate production path for API/Cloud; creator surfaces feel length/edit cost next. Watch external benches，SynthID/abuse，ad-pipeline share vs Kling，and longer continuous shots.

【Verifiable】

Primary: blog.google introducing-gemini-omni-1-2-flash (Nov 05, 2026); DeepMind + Cloud mirrors; Gemini API changelog Nov 5 GA for gemini-omni-1.2-flash.

Site: no Omni 1.2 Flash event; do not conflate with Genie 4 (e/1061). max e/1089.