Flow monthly credits by Google AI plan: how many clips fit
Flow gives 200, 1,000, 10,000 or 25,000 bonus credits a month by plan. What that buys in Veo 3.1 and Gemini Omni Flash clips, and how a Sume balance differs.
Google AI Plus adds 200 Flow credits a month, Pro adds 1,000, Ultra at $100 adds 10,000 and Ultra at $200 adds 25,000, per Google's credits page. At the published per-clip prices that is 13 to 1,666 Omni Flash clips at the top of the 15-credit range, or 2 to 250 Veo 3.1 Quality clips, depending on plan.
The plan bonuses and per-generation costs are from Manage your Google Flow credits, read 2026-10-02. Omni Flash length limits come from the Omni Flash API page and Veo's from the Veo 3.1 page.
How many clips does each plan make?
The table divides the monthly bonus by the published per-clip price. Omni Flash 720p is quoted at its top-of-range 15 credits so the result is a floor, and Veo clips are 8 seconds. Editing is 40 credits per edit.
| Plan (monthly bonus) | Omni 720p at 15 | Omni edits at 40 | Veo Lite (10 / 5 Ultra) | Veo Fast (20 / 10 Ultra) | Veo Quality (100) |
|---|---|---|---|---|---|
| Plus (200) | 13 | 5 | 20 | 10 | 2 |
| Pro (1,000) | 66 | 25 | 100 | 50 | 10 |
| Ultra $100 (10,000) | 666 | 250 | 2,000 | 1,000 | 100 |
| Ultra $200 (25,000) | 1,666 | 625 | 5,000 | 2,500 | 250 |
What else affects the count?
Free users get 50 credits a day, and the page says daily credits refresh after first use and unused amounts do not carry over. Monthly subscription credits refresh at the start of the billing cycle. Google also notes that separate AI credits may expire after a period, unlike monthly Flow credits, and that top-ups are available for Plus, Pro and Ultra but not in Japan.
The page prints a range for Omni Flash by length (7 to 15 credits at 720p, 4 to 7 at 360p), so shorter clips go further than the table's floor. 1080p upscaling is free for subscribers and 4K costs 50 credits on Ultra.
How is a Sume balance different?
Sume works from a spendable USD balance for the workspace. A submit that cannot reserve its estimated cost fails with 402 insufficient_credits before provider work starts, as the generation admission docs describe. The credit unit is therefore dollars you can read off the job, not a monthly pool that resets.
That suits automation: a script can check the estimated cost, submit with an Idempotency-Key, and poll the job. It does not suit someone who wants a bundled monthly allowance and a visual editor, which is what Flow's plans are built for.
How should you size a plan?
Count the clips you actually finish, not the ones you try. A typical selection loop makes several drafts for each keeper, so divide the table by your retry rate. If you retry five times per keeper, Pro's 66 Omni clips become about 11 finished clips.
Use cheaper drafts first. Omni Flash at 360p costs 4 to 7 credits in Flow and $0.0375 a second on Sume, and the model's 3 to 10 second range applies to both. Approve the composition at 360p, then rerun the winner at 720p or higher.
Which model should the credits go to?
Per credit, Veo 3.1 Lite is the cheapest way to get an 8-second clip in the table, and Omni Flash 360p is cheapest for a short draft. Omni Flash also offers editing at 40 credits and clips up to 10 seconds, and it is the model Google names as the Veo replacement in the API on October 22.
So a sensible split for a Pro account is cheap Lite or Omni drafts for most credits, then a few Fast or Omni 720p finals. Keep a reserve for edits, because an edit costs the same 40 credits as two Veo Fast clips.
What does this not tell you?
It does not price a Google plan in dollars, because I did not fetch the plan price pages, so dollars per credit are unknown here. It also assumes your account can pick every model listed; availability by country and plan is a separate Google page.
Sources
Related posts
More in Use cases
- Gift guide video from 20 product stills: Timeline plan and render
Hold 20 product stills for 3 seconds each over one voiceover in Timeline 1.0. Plan it free, expect a chunked render past 12 slots, and cap fades at 8 in a row.
- GPT Image 2.5 ad headline text: quote the copy, test medium vs high
For an ad still with a headline, quote the exact words, state position and type style, and render at medium and high to compare. A Sume script and its cost.
- Ad still from a packshot: tell GPT Image 2.5 what must not move
Turn a product packshot into an ad still with GPT Image 2.5 by listing what must not move: shape, colour, label text. A Sume request and a checklist.
- GPT Image 2.5 drafts at low quality, then a final: cost and steps
fal lists GPT Image 2.5 at $0.00588 low and $0.05268 high for 1024 squares. Draft four options at low, pick one, then redo it at high. Sume steps and the catch.
Written by Sume