Gemini Omni 1.1 Flash vs Gemini Omni Flash: Is the 40-Second Upgrade Worth It?

If you already built on Gemini Omni Flash, the June public preview, you have a genuine decision in front of you: Gemini Omni 1.1 Flash launched on August 27, 2026 at the same $0.10-per-second price for 720p output, and it is not a trivial bump. The preview that went out on June 30 generated ten-second clips, referenced only the final second of context when extending, capped output at 720p, and — in the small print — did not actually process the video references its own API schema accepted. The 1.1 update fixes the length problem, the resolution problem, and the reference problem in one release, without changing the price you already budgeted around. Whether that means you should move, and how fast, depends on which of those gaps you were actually hitting.

The family relationship matters here. Gemini Omni Flash is Google's multimodal video generation and conversational-editing model — text, image, and video inputs in, short clips with synchronized audio out — aimed at iterative, language-driven editing rather than one-shot generation, which is the job Google reserves for Veo 3.1. Gemini Omni 1.1 Flash is the same product line's production update, and it inherits the whole interaction model: same conversational multi-turn editing, same any-input design, same SynthID watermarking. The differences are concentrated in four specific controls, and they are worth enumerating one by one because the version number undersells them.

What the 1.1 update changes

• Scene extension is the flagship change. Gemini Omni Flash generated a single 10-second clip; Gemini Omni 1.1 Flash analyzes up to 10 seconds of prior context — a leap from the earlier model's one-second lookback — and extends footage in 10-second increments to a 40-second cumulative ceiling. For anyone producing anything longer than a moment, this is the difference between a generator and a tool.

• Resolution ladder. The preview output native 720p and stopped there. Gemini Omni 1.1 Flash renders at 720p as standard but also supports 1080p and 4K upscaled output, which moves the model from social-native to broadcast-plausible. Per-second pricing for the high-resolution tiers is not published yet — only the 720p $0.10/s figure is official.

• First/last frame control is new. You can pin the opening and closing frames of a shot and have the model generate the continuous motion between them. Camera orbits, zoom transitions, and seamless loops all become prompt constraints instead of lucky outcomes.

• 360p drafting is a new cost tier. Previews render up to 60% faster than 720p at roughly a third of the cost, per Google's vendor-reported throughput figures — a deliberate iteration lever for storyboarding before committing to a full render.

• Video references work now. Gemini Omni Flash's API accepted up to three seconds of reference video but did not correctly process it at launch. Gemini Omni 1.1 Flash treats up to three seconds of reference video as a real multimodal input, which is what makes character and visual consistency across shots plausible instead of promised.

What did not change

It is just as important to know what stayed put, because "upgrade" implies more than Google actually shipped. The 720p price is identical: $0.10 per second of video output, $1.50 per million input tokens, $9.00 per million text output tokens — the structure carried over untouched. The core editing model is the same conversational, multi-turn loop, so workflows you built against the preview's interaction pattern keep working. And the honest gaps remain honest: character consistency across scene changes is improved by the longer lookback but still a generative weakness to test, audio references are still not part of the API contract, and there is still no independent public benchmark for either version — Google's quality and speed claims are vendor-reported, full stop.

Should you switch?

The answer breaks into three buckets. If you hit the 10-second wall — you needed scenes longer than a moment and were stitching clips in post — Gemini Omni 1.1 Flash is not an upgrade you can defer; it is the fix to the exact bug you were working around, at the same price. If you produce for anything above social resolution, the 1080p/4K output settles it as soon as Google publishes the high-res per-second rates, because until then the budget is an estimate. And if you were staying on the preview because of cost, the 360p drafting tier is the reason to move now: iterating at a third of the price changes how many drafts a shot is worth, and that arithmetic compounds across a production calendar.

The migration itself is small but not automatic. The model ID changes — code>gemini-omni-flash-preview/code> to code>gemini-omni-1.1-flash/code> — and the new capabilities are enabled by request fields like code>resolution/code> in the interaction call, so your code gains options rather than losing behavior. If you want to run old and new side by side while you validate, that is a configuration decision, not a rewrite.

Where the cost discipline lands

The quiet theme of this release is that Google is finally pricing iteration as a first-class feature — a 360p tier that costs a third as much exists to be used wastefully, and that is the point. The same pass-through philosophy is what a routing layer is for. OrcaRouter passes provider list prices through with zero markup, so the $0.10/s 720p rate and the 360p discount show up on our side at Google's number, not a marked-up one, and when a vendor cuts a price or adds a tier, the change is live the same day — one API for 200+ models, with automatic failover if a provider hiccups mid-render. Gemini Omni 1.1 Flash is not routable through us yet — Google has not opened the Omni API to third-party routing, and we do not claim to host what we do not — but the mechanism is the same one that already carries Google's image and text models at list price. When the Omni API widens, it lands in the same catalog at the same discipline.

Bottom line

Gemini Omni 1.1 Flash is what Gemini Omni Flash should have been at launch: 40-second scenes with a real lookback, first/last frame control, 1080p/4K output, working video references, and a 360p drafting tier at a third of the cost — all at the same $0.10/s 720p price. The upgrade decision only has teeth because the price did not move; you are getting the length, resolution, and reference fixes for free relative to the preview budget. The remaining unknowns are Google's unpublished 1080p/4K rates and the total absence of an independent benchmark. Switch if you hit the length wall, the resolution wall, or the iteration-cost wall. If none of those are hurting you, the preview still works — but the gap between the two versions is exactly the gap between a demo and a deliverable.