Omni Flash Pricing 2026: Flow Credits & Real Cost Guide

I spent the first weekend after I/O running Omni Flash through Flow to figure out one thing: what does omni flash pricing actually look like once you stop counting “videos per month” and start counting credits per generation. The marketing page is clean. The math is not.

If you’re deciding whether to upgrade a subscription, or you’re the person at a small team putting a number next to “AI video” in next quarter’s budget, this is the breakdown I wish I’d had on day one. I’m Dora.

How Omni Flash Is Currently Priced

No standalone per-second rate yet

Omni Flash does not have a public per-second API rate. Google launched the model on May 19, 2026, said API access is coming “in the coming weeks,” and has not committed to a number. The real question isn’t “what does Omni Flash cost per second” — that question has no answer right now. It’s “how much does the subscription cost, and how far do the credits go.”

Embedded in Google AI subscription + Flow credit system

Omni Flash sits inside Google’s AI subscription tiers, accessed through the Gemini app or Flow. You don’t buy Omni Flash. You buy Plus, Pro, or Ultra. Flow credits are the usage currency.

Subscription Tiers That Unlock Omni Flash

Tier	Monthly Price	Monthly Flow Credits	Best For
Google AI Plus	~$10–13 (region-dependent)	200	Light testing
Google AI Pro	$19.99	1,000	Solo creators, testing
Google AI Ultra	$199.99–$249.99	10,000 or 25,000	Studios, agencies

A note on Ultra: gagadget reported Google trimmed Ultra from $250 to $200 a month around launch, part of an aggressive distribution play. So the Ultra number depends on when the page was cached. Confirm before subscribing.

The number that matters most for google ai pro omni flash users: 1,000 Flow credits per month. Everything else gets measured against that line.

The free Gemini tier and YouTube Shorts integration give you Omni Flash access without paying, capped at roughly 50 daily credits on Flow — enough for one or two short generations a day, as of publication.

Flow Credits per Generation

This is the part most cost estimates skip. Based on credit consumption observed in Flow during the first week:

Output	Flow Credits
4-second clip	15
6-second clip	20
8-second clip	25
10-second clip	30
Edit existing video	40

The 10-second cap is a hard ceiling. Nicole Brichtova of Google DeepMind told TechCrunch the limit is a deployment choice, not a model constraint — a way to manage compute demand. You can’t pay more credits for a 30-second clip from Omni Flash. You can only generate more 10-second clips and stitch them.

Editing costs more than generating. 40 credits for an edit pass versus 15 for a fresh 4-second clip. Counterintuitive, but it makes sense: an edit takes your existing video as input, reprocesses every frame, maintains consistency with what’s already there, and runs the full generation pipeline on top. A new 4-second clip starts from text. An edit starts from video — strictly more compute. Practical consequence: regenerating four takes is often cheaper than generating one and editing it twice.

What a Real Workflow Actually Costs

Credit numbers don’t mean much until you map them to volume.

Scenario 1 — Solo creator (100 videos/month). Mostly 6-second clips for social. 100 × 20 = 2,000 credits. Pro’s 1,000 covers about 50 of those. Pro doesn’t fit a solo creator running heavy social volume at 6+ seconds.

Scenario 2 — Small marketing team (500 videos/month). Mostly 8-second clips: 500 × 25 = 12,500 credits/month. Pro is irrelevant at this volume. Ultra’s 25,000 tier is the realistic floor. Split across a 5-person team: 5,000 credits each, about 200 eight-second clips per person.

Scenario 3 — Product team prototyping. Different shape. 30 final clips a month, but each takes 4–6 attempts plus 2–3 edits to land. Per final clip: roughly (5 × 25) + (2 × 40) = 205 credits. Across 30 finals: about 6,150 credits. Pro can’t cover it. Ultra 10,000 has comfortable headroom.

The pattern across all three: Pro fits low-volume creators or testing. Anything resembling production volume jumps straight to Ultra. No middle.

Hidden Cost Drivers Most Estimates Miss

The “100 clips × 20 credits = 2,000 credits” math is the headline. The real number is higher.

Re-generations on failed outputs. Safety filters fire on prompts they shouldn’t. A Google AI Studio forum post from early in the rollout documents a 30-credit charge on a clip flagged as potentially harmful and never delivered. The credits were not refunded. Factor in 5–15% wastage from false positives.

Iterative editing cycles. Almost no first generation is the final generation. If your workflow has a quality bar, you’ll regenerate and edit. The 40-credit edit cost compounds fast.

Multiple takes for consistency. Multi-shot sequences need character and lighting consistency across clips. Getting two shots of the same character to match often takes 3–5 attempts per shot.

A rough adjustment: take your clean math and multiply by 1.5–2x for real consumption.

API Pricing — What We Know and Don’t

Google hasn’t published it. The Omni Flash model card on DeepMind’s site describes inputs and outputs, but pricing is absent. The official Flow announcement post says API access “is rolling out in the coming weeks.” If you’re planning a product around Omni Flash API, you cannot model unit economics yet.

Based on how Veo 3 and Gemini Flash variants are priced today on Vertex AI, an educated guess for Omni Flash sits in the $0.20–$0.60 per second of video output range. This is a guess. Don’t build a business case on it.

When Subscription Is Right vs When to Wait for API

Subscription path makes sense when you generate clips inside a creator workflow, your monthly volume fits the credit allocation with room for waste, and you want predictable cost.

Wait for API when you’re integrating into a product and need per-clip cost predictability, your usage is spiky, or subscription credits don’t map to your customer-facing pricing.

For most readers in May 2026: Pro at $19.99 if you’re solo or testing, or wait until API rates publish if you need integration.

FAQ

Can I use Omni Flash on the free Gemini tier?

Yes, with limits. The free path runs through Flow (around 50 daily credits as of publication) or YouTube Shorts and YouTube Create at no cost. Enough to test. Not enough for sustained omni flash credits workflow.

Do Flow credits roll over to the next month?

Google has not publicly clarified rollover behavior in its current subscription terms. Industry default is no rollover — credits expire at month-end. Treat that as the working assumption. Check the latest subscription terms before subscribing — this needs verification.

Why does editing a video cost more credits than generating a new one?

Edits reprocess your existing video frame-by-frame, maintain consistency with the parts you didn’t change, and run the full generation pipeline on top. That’s strictly more compute than generating fresh from text. Practical takeaway: regenerate variations rather than edit incrementally when budget matters.

Has Google announced Vertex AI API pricing for Omni Flash?

Not as of May 2026. Google said API access is “coming in the coming weeks” at I/O, but no per-second or per-token rate has been published. The current omni flash cost discussion is subscription-only.

Are there enterprise pricing or volume discounts available?

Not publicly documented. Google AI Ultra is the highest published tier. Anything beyond — volume discounts, enterprise contracts, dedicated quota — would go through Google sales. Needs verification with a Google account rep.

Is Google AI Pro enough for a 3-5 person marketing team?

Run the math. Pro’s 1,000 credits split across 5 people is 200 credits each — roughly 10 eight-second clips per person per month. That’s testing, not production. For 3–5 person marketing teams running real volume, google flow credits omni flash allocations on Pro fall short, and Ultra at the 25,000-credit tier is the realistic plan. The omni flash subscription cost jumps significantly at that point, but the per-clip math finally works.

Bottom Line

Omni flash pricing in May 2026 is two questions: how much does the subscription cost, and how many credits does your real workflow burn through. Pro at $19.99 with 1,000 credits is the entry point for solo creators and testing. Production teams jump to Ultra. The API rate isn’t public yet, which means product integrations should run on subscription tiers and wait for the rate card before committing to architecture.

I’ll re-run the credit math against my own logs in a month and see whether the 1.5x wastage adjustment holds. This is where my data ends for now.

Previous posts：