By the end of this you'll know every legitimate way to generate a Veo 3.1 clip in 2026 and which one fits your work. Veo 3.1 is Google DeepMind's flagship AI video model, producing native 4K/60fps clips with synchronized audio from a text or image prompt. If you want the model's full capability breakdown, what is Veo 3 covers it. This is the access guide: the official Google routes, the per-clip canvas route, and a straight decision table so you're not guessing.
The short version on cost: a finished 5-second Veo clip runs anywhere from a slice of a $20 subscription to about $3.75 on the raw API to $0.85 to $1.20 per clip on a canvas. Which number applies is entirely a function of how you access the model, so let's walk through each door.
The official Google routes
Google sells Veo access three ways, and all three are legitimate first-party options.
The Gemini app. The consumer route. A Gemini subscription unlocks Veo generation inside the Gemini app and Google Flow. The free tier gives a small monthly credit allowance for limited Veo access. Google AI Pro, around $19.99 a month, unlocks Veo 3.1 with roughly 1,000 monthly AI credits. Google AI Ultra, around $249.99 a month, adds a much larger credit pool and priority access to the newest versions. This is the simplest route if you already want a Gemini subscription for other reasons.
Google Flow. Google's dedicated creator front-end for Veo, priced per generation in credits: roughly 10 credits for a Lite generation, 20 for Fast, and 100 for the full Quality preset. Non-subscribers get a small pool of free daily credits to test. Flow is where the Gemini-tier credits actually get spent on video, and it's the most feature-complete first-party interface for the model.
Vertex AI. The developer route, billing per second of generated video at around $0.75 per second for the Standard tier through your Google Cloud account. A standard 8-second clip runs about $6.00. Cheaper Gemini API variants exist for lower tiers (Fast around $0.10 per second, Lite around $0.05). This route is for building Veo into your own software, not for making individual videos by hand.
These routes are real, first-party, and the right call in specific situations covered in the decision table below. The point of the canvas route that follows isn't that Google's options don't exist. It's that Veo rarely gets used alone, and per-clip access on a multi-model canvas fits real production better when your volume is uneven and your work chains across models. For the full pricing breakdown of all four routes, see Veo 3.1 pricing explained.
The 8frame canvas route, step by step
Here's how to generate a Veo 3.1 clip on the 8frame canvas, where the model sits next to 15 others and one credit pool covers them all.
Step 1: Open the canvas and pick Veo 3.1
Go to app.8frame.co and open a canvas. Veo 3.1 sits at the top of the video model roster, the default pick when cinematic look or native audio is the requirement. Select it as the model for your generation node. Plans start at $19/month for 1,000 credits; full Veo 3.1 runs 112 credits per 8-second clip, Veo 3.1 Fast is 56, and Veo 3.1 Lite starts at 17.
Step 2: Add references if you're doing image-to-video
If you have a starting still, a product shot, a brand frame, a reference for composition, drop it onto the canvas and wire it into the Veo node. Veo does image-to-video as well as text-to-video, and conditioning on a reference gives you tighter control over the look than a text prompt alone. For a reference still, generate one first on Nano Banana Pro on the same canvas and chain it straight in.
Step 3: Write the prompt with lighting and audio direction
Veo responds precisely to lighting, lens, and audio language, so use it. A prompt like "Rooftop at dusk, city lights coming on, condensation on a glass bottle in the foreground, shallow depth of field, wind, ambient city noise" gets you the color transition, the bokeh, and a matched ambient track in one generation. The native audio is a real advantage: describe the sound and Veo syncs it rather than making you route through a separate audio model. The Veo 3 prompt guide has the full prompt structures.
Step 4: Generate, review, and chain the winners
Generation takes about 90 seconds and returns a native 4K/60fps clip. Review it on the canvas. Because everything lives on one surface, you can draft cheaper variants on Kling first and only run the winning shot on Veo, then chain the approved clip into a Topaz upscale or an edit node without leaving the canvas. That draft-cheap, finish-on-Veo move is the single biggest lever on your Veo spend, and it only works when the models share a surface.
Which route to pick
| Your situation | Best route |
|---|---|
| Already paying for Gemini Pro/Ultra | Gemini app / Flow (bundled) |
| Building Veo into your own product | Vertex AI (per-second API) |
| Very high, steady Veo-only volume | Gemini Ultra (large credit pool) |
| Uneven volume, credits that roll over | 8frame (from $19/mo) |
| Chaining Veo with other models | 8frame (multi-model canvas) |
| Watermark-free clips, all models on one plan | 8frame |
| Testing the model for free | Gemini free tier or Flow free credits |
The honest split: if Veo is bundled into a Gemini subscription you already pay for, or you're a developer who needs programmatic API access, Google's own routes are correct and no canvas replaces them. If your volume is uneven, you route work across several models, or you want one plan instead of a stack of them, the canvas route wins.
FAQ
Can I use Veo 3.1 without a Gemini subscription?
Yes. The Vertex AI API bills per second with no subscription, and the 8frame canvas offers Veo 3.1 at 112 credits per clip (Lite from 17) on one $19/month plan that also covers 15 other models, watermark-free. A Gemini subscription is only one of several legitimate routes to the model, not a requirement.
What's the cheapest way to generate a Veo 3.1 clip?
For low, uneven volume, a canvas credit pool is usually cheapest because unused credits roll over and you can draft on Veo 3.1 Lite from 17 credits before finishing on the full model. For very high steady volume, a Gemini Ultra subscription's large credit pool can win. The raw Vertex API at $0.75 per second is the most expensive per finished clip unless you're at negotiated enterprise scale.
Does Veo 3.1 generate audio?
Yes. Native synchronized audio is one of Veo's defining features. Ambient sound, dialogue, and music are generated together with the visuals and sync to what's on screen, so you don't route the output through a separate audio model. Describe the sound in the prompt and Veo produces it alongside the picture.
Can I use Veo 3.1 for commercial work?
Yes, with model-specific licensing. Veo output on paid routes, including 8frame on any plan, permits commercial use. Free evaluation allowances in Google's own products have their own terms, so check the licensing before shipping a free clip into a paid deliverable.
Generate Veo 3.1 alongside 15 other models from one credit pool, from $19/month. Open the 8frame canvas and run your first Veo clip today.