Google Veo Pricing Explained (2026): Per-Second API Costs, Free Tier, and What a Real Video Costs

Photo: Sudhakar Chandra, Unsplash License.

Veo pricing is genuinely confusing because Google sells it two different ways — a metered API rate and a flat-fee consumer subscription that bundles credits — and the cheapest option depends entirely on how much video you actually generate per month. Here’s the real math.

The short answer

• API pricing (Veo 3.1, per second, with audio by default): Standard $0.40/s (720p/1080p) or $0.60/s (4K); Fast $0.10-0.12/s (720p/1080p) or $0.30/s (4K); Lite $0.05-0.08/s (no separate 4K tier).
• Consumer subscriptions: Google AI Pro at $19.99/month and Google AI Ultra at $249.99/month both bundle monthly AI credits usable for Veo generation instead of metered per-second billing.
• Cheapest path for occasional use: the Pro subscription’s bundled credits, if your monthly video volume is low. Cheapest path for high-volume/automated pipelines: the metered API on the Fast or Lite tier, where cost scales predictably with usage.

Per-second API pricing, in full

Tier	Resolution	Price/second	Notes
Lite	720p	$0.05	Cheapest generation tier
Lite	1080p	$0.08
Fast	720p	$0.10
Fast	1080p	$0.12
Fast	4K	$0.30
Standard	720p/1080p	$0.40	Highest fidelity below 4K
Standard	4K	$0.60	Top tier

These rates include audio generation by default on Veo 3.1 — Google’s own documentation notes you’re only billed for videos that actually finish generating successfully, so a failed audio-processing pass on a given clip isn’t charged.

What a real clip costs

An 8-second b-roll clip (a common single-shot length for narrated long-form video) costs:

• Lite 720p: 8 × $0.05 = $0.40
• Fast 1080p: 8 × $0.12 = $0.96
• Standard 1080p: 8 × $0.40 = $3.20

A 10-minute narrated video built from roughly 40-50 such 8-second clips (a rough estimate — actual clip count depends heavily on pacing and how much you mix in static images) lands very differently depending on tier: roughly $16-20 on Lite, $38-48 on Fast, and $128-160 on Standard for video generation alone, before narration, editing, or any other production cost. This is exactly why picking the right tier for the job — Lite/Fast for disposable b-roll, Standard only for hero shots — matters more with Veo than with a flat-fee tool.

Subscription vs. metered API: which is actually cheaper

Google AI Pro ($19.99/mo) bundles a fixed monthly credit allowance that covers a limited amount of Veo generation alongside Gemini’s other features (chat, image generation, etc.) — it’s a good fit if your video generation is occasional and you’re not running an automated pipeline that needs programmatic API access.

Google AI Ultra ($249.99/mo) raises that credit ceiling substantially and is aimed at heavier individual creators, not automated production pipelines — at that price point, most channels producing more than a handful of videos a month will find the metered API cheaper and more predictable, since Ultra’s cost doesn’t scale down when you generate less.

The metered Gemini API is the only option that supports direct programmatic/automated generation (as opposed to the consumer app UI), and its cost scales linearly with actual usage — the right choice once you’re running Veo through a script or batch pipeline rather than generating clips one at a time by hand.

Cost-control tips for a creator pipeline

Default to Lite or Fast for b-roll, reserve Standard for hero shots. Most seconds of a finished video are supporting footage, not the single shot the thumbnail is built around — there’s rarely a visible quality difference for background motion at Lite/Fast tiers.

Don’t regenerate a whole clip to fix a small issue. Since you’re billed per successfully generated second, iterating with shorter test generations (a 2-3 second preview) before committing to a full 8-second clip at your final tier keeps failed attempts cheap.

Track cost per finished minute, not per clip. Because clip length, tier, and resolution all move the price independently, the only stable metric across a whole production is total spend divided by finished video minutes — track that per project so tier choices you make now don’t quietly blow the budget on a longer video later.

Mix static AI images with generated video. A video doesn’t need every second to be Veo-generated motion — cutting in still images (from Midjourney or Leonardo AI) for lower-motion sections cuts the total seconds of video generation you’re billed for. See Midjourney vs Leonardo AI for that side of the pipeline.

FAQ

Is there a free tier for Veo? Not on the metered API — Google AI Pro’s bundled monthly credits are the closest thing to a low-cost entry point, and even those are consumed by other Gemini features, not reserved solely for video.

Does 4K always cost more than it’s worth for YouTube? For most YouTube delivery, 1080p is the practical ceiling most viewers experience — the 4K tier’s roughly 50% price premium over 1080p at the same speed tier is worth it mainly for hero shots you expect viewers to pause on, not for full-video generation.

How does Veo pricing compare to a subscription image/video tool like Leonardo AI? They’re structured completely differently — Leonardo’s token system is a flat monthly allowance regardless of how much of it you use, while Veo’s API bills exactly for what you generate. A high-volume, automated pipeline usually favors Veo’s linear API pricing; an individual creator generating casually often prefers a flat subscription’s predictability.

Did older Veo 3.0 pricing carry over? No — the legacy veo-3.0-generate-001 and veo-3.0-fast-generate-001 endpoints were retired mid-2026, so any new integration should be built against Veo 3.1’s current pricing above, not older published rates.