Gemini Omni Flash 1.1 adds video references and 10-second extensions

• Video can become a 40-second chain: OfficialLoganK's launch post says Omni 1.1 reads up to 10 seconds of prior footage before each 10-second extension, while Google's developer guide sets the 40-second cumulative limit.
• Camera moves can be designed from two endpoints: chrisfirst's feature list confirms first- and last-frame support, and Google says the model generates the continuous shot between them.
• Video can now act as source material. Pika API Club's feature post says Omni accepts up to three reference videos; Google's guide demonstrates a character-and-choreography transfer across three dance clips.
• The resolution path now separates iteration from finish: OfficialLoganK's thread identifies 360p drafts and 4K upsamplers, while Google puts 360p at up to 60% faster and roughly one-third the cost of 720p.
• The rollout reaches Google surfaces and a third-party API route, as the Pika API Club announcement offers Gemini Omni 1.1 through its catalog.

Google's developer guide includes a no-cut prompt where dog, octopus, and bear characters inherit three different dancers' choreography. The Flow update makes the low-credit draft loop mobile: create at 360p, then download a favored version at 720p.

Omni 1.1 extends clips in 10-second increments and uses up to 10 seconds of earlier footage as context, replacing the previous final-second reference window. Google's developer guide caps the accumulated sequence at 40 seconds.

At the API layer, an extension supplies a previous_interaction_id and a new text instruction. Google's examples use that continuation step for a dolly zoom, a snap zoom, and a 360-degree orbit around a frozen character.

First and last frames act as the boundary images for a generated continuous shot. Google positions the feature for camera orbits, zoom transitions, and seamless loops in its developer guide.

The update accepts video alongside images and text as multimodal reference material. Pika API Club's feature post describes up to three reference videos, while Google's published example assigns a separate uploaded dance to each supplied character within one uninterrupted shot.

Google's feature description calls for up to three seconds of reference footage.

Google separates quick exploration from final delivery with three resolution stages:

• 360p: preview renders, up to 60% faster and about one-third the cost of standard 720p, according to Google's launch post.
• 720p: the standard comparison render.
• 1080p or 4K: polished outputs intended for production delivery.

Google splits access by surface in its rollout note: AI Studio exposes the Gemini API, the Enterprise Agent Platform exposes its API, Flow gets the full set for Google AI Plus, Pro, and Ultra subscribers, and the Gemini app gets scene extension.

Pika's Gemini Omni 1.1 Flash text-to-video documentation lists 360p, 720p, and 1080p outputs in 16:9 or 9:16. Its API returns a queued job, then requires polling for a completed or failed result.