FULL STORY

Gemini Omni 1.1 Flash: Leak to Leaderboard Top

Google's unreleased Gemini Omni 1.1 Flash surfaced in Cloud Console before its official launch as a video generation model, then topped the text-to-video Video Arena leaderboard.

2026-08-27 ~ 2026-08-29 · 3 episodes · 35 posts

Episode 1 · Unreleased Gemini Omni 1.1 Flash Appears in Google Cloud (2026-08-27, 2 posts)

Reddit users spotted an unreleased Google Gemini "Omni 1.1 Flash" model in Google Cloud's console, docs, and quota settings, suggesting it may launch within 24 hours.

Episode 2 · Google Launches Gemini Omni 1.1 Flash with 4K Output and Scene Extension (2026-08-27, 29 posts)

Google officially launched Gemini Omni 1.1 Flash (internally codenamed Omni 1.2 according to @legitapi) on August 27-28, making it available to developers via the Gemini API and Google AI Studio, with Flow and the Gemini App updated in parallel. The release integrates Veo-style creative controls, with the headline upgrade being scene extension based on up to 10 seconds of context. Google DeepMind said the model is designed to improve controllability, speed up iteration, and fit production use cases.

Confirmed

  • Official (@GoogleAI) announcements confirm Veo creative controls, 4K upsampling, first/last-frame control, fast 360p drafts, and scene extension based on up to 10 seconds of context, plus Flow and Gemini App updates
  • Scene extension analyzes up to 10 seconds of prior footage to continue a scene while preserving character identity, lighting, and narrative context; scenes can extend to 40 seconds in 10-second increments; video understanding also extends to 40 seconds with keyframe (first/last frame) extraction (@osanseviero)
  • Controls include first/last-frame specification, 3-second reference videos, frame interpolation, motion reference, and multiple video references (@legitapi, @philschmid)
  • Resolutions and cost: 1080p and 4K output, with a low-cost 360p draft mode for fast iteration before upscaling; @DeryaTR relayed user reports of stunning 4K upsampling without complex prompts
  • The Gemini App team (@GeminiApp) opened Scene Extension to all Google AI Plus, Pro, and Ultra subscribers worldwide via the Create video tool; @jasonbaldridge noted Flow desktop gets first/last-frame control and 4K export, mobile gets 360p draft mode
  • @philschmid noted cleaner motion in this update; @OfficialLoganK described it as an "everything in, everything out" world model

Why it matters

  • This Flash-tier iteration of the Omni family brings Veo-grade creative control to a cheaper model, and the 360p-draft-then-4K-upscale workflow substantially cuts the cost of iteration in video generation
  • Character consistency and 10-second-increment scene extension up to 40 seconds target long-form narrative video production, a key competitive capability against rivals
  • Opening Scene Extension to all paid Google AI subscribers moves the capability from developer APIs into ordinary consumers' creative workflows

9 more related posts →

Episode 3 · Gemini Omni 1.1 Flash Tops Text-to-Video Arena Rankings (2026-08-28, 4 posts)

Google's updated Gemini Omni 1.1 Flash has claimed first place in text-to-video and second in image-to-video (1488 points) on the Video Arena leaderboard. The update shifts video generation toward more controllable, director-style creation with features like scene extension.