Correction, 2026-10-02: this page said Omni works out what a scene should sound like "so footsteps land on footfalls and a room's ambience matches the room". We have not measured that; the page now gives Google's own description and its advice to describe the sound in the prompt. It also called Omni Flash "the variant developers and platforms actually use" (Google calls it the first model in the Omni family), said Google AI subscriptions come with "monthly generation allowances" for Omni (Google's page says only that a subscription is required and that features vary by tier and geography), headed a section "Reference-driven consistency", which we have not tested either, and closed by saying 8frame runs "every other major model", which it does not. We now also say that per clip Google's API is cheaper than 8frame, and list what the 8frame tool does not do: no end frame, no extension, no 360p, 1080p or 4K.
Correction, 2026-10-01: we called three-second Omni passes the cheapest way we had found to test a beat with sound. They are not the cheapest clip with sound on 8frame (Creatify Boreal is 7 credits for 5 seconds at 720p, Grok Imagine 34 at 480p). We also now say that on 8frame Omni Flash returns 720p and makes 3 to 10 second clips; the 1080p and 4K upscales and the extension to 40 seconds are on Google's own API.
Correction, 2026-09-30: we quoted our own production volumes on this page, which we do not publish, and said earlier Omni models referenced the final frame when extending; Google says the final second. Both fixed.
Gemini Omni is Google's multimodal model line for video, and its defining trait is that audio is generated with the picture rather than added afterward.
Gemini Omni Flash is the first model in the family, and the one Google offers in the Gemini app and on its Gemini API. Google announced it on 2026-05-19 and brought it to the Gemini API and AI Studio in public preview on 2026-06-30; Gemini Omni 1.1 Flash, released on 2026-08-27, is the generally available version. In the Gemini app, Google says "Gemini Omni will replace Veo". Google prices video output on its own API at $17.50 per 1M output tokens, which its pricing page puts at roughly $0.10 per second of 720p video. On 8frame an 8-second clip with sound is 135 credits, about $1.35 at pack rate.
How Gemini Omni works
The model takes text, images and video as input and returns a clip with audio. Google's documentation says the model "will try to generate an appropriate audio track" by default and that you can describe the audio you want in the prompt, which matters most if you want music. So write the sound into the prompt the way you write the shot: "calm background music", "rain on the window, no dialogue". On 8frame there is no audio toggle: every Omni clip arrives with sound.
Google's Omni Flash documentation lists the model as gemini-omni-1.1-flash, with 16:9 (the default) and 9:16 aspect ratios, 360p and 720p (the default) output, and 1080p and 4K as upscales rather than native renders. On Google's API, clips can be extended in 10-second steps to 40 seconds in total, and an uploaded video for editing or extending is capped at 10 seconds. One regional catch on Google's API: editing or extending an uploaded video is not available in the European Economic Area, Switzerland or the UK, though editing or extending a video the model generated is.
The same guide carries a prompting note worth knowing before your first clip: by default Omni Flash "will try to create a video with a few different shots". If you need one unbroken shot, say so in the prompt, for example "in a single continuous shot".
Google announced the 1.1 generation on 2026-08-27, adding scene extension, first and last frame control, and up to 10 seconds of prior context when extending, against earlier models that referenced only the final second. It also added 360p drafts, which Google puts at up to 60% faster and a third of the cost of 720p.
Gemini Omni on 8frame
Not all of those API features are in the 8frame tool. What it does:
- Length: 3 to 10 seconds per clip, in whole seconds. A new node opens at 5 seconds (85 credits).
- Shape: 16:9 (the default) or 9:16.
- Resolution: 720p. There is no resolution setting, so no 360p drafts and no 1080p or 4K.
- Sound: always on.
- Modes: picked from what you connect. A prompt alone runs text-to-video, a start image runs image-to-video, one to seven reference images run reference-to-video (a start image and reference images can't be combined), and an attached clip of 3 to 10 seconds runs video edit, which keeps the clip's length and is priced by it.
- Not in the 8frame tool: an end frame and clip extension.
- Speed: in our production data, 3- to 5-second clips took a median of about 35 seconds and video edits about 49 seconds. We have no measured figure for 8-second clips.
What it costs where
| Route | Price | Notes |
|---|---|---|
| Gemini API | About $0.10 per second at 720p, billed as $17.50 per 1M output tokens | Google's published rate; no free tier |
| Gemini app | Google AI subscription required | "Features vary by tier and geography"; at launch Google also rolled Omni Flash out to Google AI Plus, Pro and Ultra subscribers in Google Flow |
| YouTube Shorts and YouTube Create | At no cost, per Google's launch post | Announced 2026-05-19 |
| 8frame canvas | 51 credits at 3s, 85 at 5s, 135 at 8s, 169 at 10s | Sound always included, 720p, no watermark |
Per clip, Google's API is cheaper: 8 seconds at 720p comes to about $0.81 there (Google bills 5,792 tokens per second of 720p video), against about $1.35 on 8frame, but the API needs a developer setup. On 8frame a credit is about $0.01 at pack rate, and the $19 Starter plan carries 1,000 credits a month, so about 7 eight-second Omni clips or 19 three-second ones. Failed generations are refunded automatically.
Examples
Drafting with sound in the loop. Because sound comes with every Omni clip, a draft lets you judge picture and sound together. A 3-second Omni pass is 51 credits, so you can check a beat with sound before paying for a 10-second clip (169). It is not the cheapest clip with sound on 8frame: Veo 3.1 Lite is 27 credits for 4 seconds with sound, at the same 720p.
Several references in one shot. Connect up to seven reference images, say a character, a product and a location, and the tool runs reference-to-video. Google's own example uses two images, a cat and some yarn. How closely the clip follows your references is something to check on your own images before you scale up.
Editing a clip you already have. Attach a 3- to 10-second clip and describe the change. The result keeps the clip's length, so editing a 5-second clip costs 85 credits. Google's guide recommends short edit prompts such as "Make this video anime", and adding "Keep everything else the same" when you change one thing.
FAQ
What does Gemini Omni Flash cost on 8frame? 51 credits for 3 seconds, 85 for 5, 135 for 8 and 169 for 10, with sound always included.
Is Gemini Omni replacing Veo? In the Gemini app, yes: Google says "Gemini Omni will replace Veo in the Gemini app." Veo 3.1 is still listed on Google's Gemini API, and on 8frame both models are available. Gemini Omni vs Veo 3.1 compares them.
Can Gemini Omni Flash make 4K on 8frame? No. It returns 720p on 8frame, and no video model on 8frame outputs 4K. Google's own API offers 1080p and 4K as upscales.
Can Gemini Omni Flash extend a clip on 8frame? No. Each clip is 3 to 10 seconds, and video edit keeps the length of the clip you attach. Extension up to 40 seconds is a feature of Google's API.
Does Gemini Omni Flash have a silent mode on 8frame? No. Every clip comes with sound, so you pay for it even if you mute it.
Sources
- Launch, preview and 1.1 dates and features: Google's Gemini Omni announcement (2026-05-19: first model in the Omni family; rollout to Google AI Plus, Pro and Ultra subscribers in the Gemini app and Google Flow, and at no cost on YouTube Shorts and the YouTube Create app), public preview post (2026-06-30), Omni 1.1 Flash post (2026-08-27: extension, first and last frames, 10 seconds of prior context, 360p drafts) and the Gemini API changelog (generally available on 2026-08-27), checked 2026-10-02
- Model spec, extension, limits, regional note and prompt guide: ai.google.dev/gemini-api/docs/omni (page last updated 2026-09-23), checked 2026-10-02
- API price and free tier: ai.google.dev/gemini-api/docs/pricing ($17.50 per 1M output video tokens, 5,792 tokens per second of 720p, about $0.10 per second; free tier "Not available"; page last updated 2026-10-01), checked 2026-10-02
- Gemini app: Gemini video generation ("Gemini Omni will replace Veo in the Gemini app"; "Google AI subscription required. Features vary by tier and geography."), checked 2026-10-02
- 8frame prices: backend tool registry (calculateGeminiOmniFlashCost), read 2026-10-02; modes, lengths, shapes, reference limit and the 5-second default: front-end tool registry, read 2026-10-02; measured 720p output and generation times: 8frame production data, read-only, 2026-09-30
Related concepts
Gemini Omni vs Veo 3.1 puts the two Google video models side by side on price, length, inputs and output spec. What is Veo 3.1 Lite covers the cheapest Veo tier, and best AI video generator with sound 2026 covers the wider field of models that generate audio natively.
Sound comes with every Omni clip, so write it into the prompt. Run Gemini Omni Flash from 51 credits on the 8frame canvas, next to Veo, Kling, Seedance and the other models there, from $19/month.
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.