Budget in credits, not USD per second
Plan the week in catalog credits on named rows. A host’s USD per second is not your invoice.
Budget this week in credits, not USD per second. The catalog quotes credits on a named row, at the options you actually select. Host blogs quote dollars per second for their API. Those dollars are not this invoice, they go stale when a host changes a cell, and they are how a producer “knows” a clip costs a host’s unit price and then spends four silent retries they never put in the sheet.
This is not a trend piece about how video is metered. Credits per second as a unit already exists. This page is the Monday habit: write a credit budget for the week’s slate before anyone generates, and refuse to translate it into a host’s unit price.
Versely does not publish USD per second. Third-party USD belongs to the third party. Do not paste it into a client estimate as if it were this studio.
The week is a pile of named rows
A useful budget is a table, not a vibe:
| This week’s job | Named row (example) | How you count | Credits you actually plan |
|---|---|---|---|
| Silent motion tests | Kling 3 Turbo T2V | 8 credits × seconds × tries | Tries, not one take |
| Short talking hero | Veo 3.1 | 64 credits on the card, 4 / 6 / 8s, doubled with sound | Hero + a small retry cap |
| Longer sounded hold | Seedance 2.0 or 2.5 | 44 or 34 on the card, duration ladder | One lock, not a shop |
| Captions | Basic 1× or dynamic 2× | Per clip, multiplier | House tier, then count clips |
| Omni revision | Gemini Omni Flash Edit | 50 credits on the card (10 a second of source), per edit turn | Turns, not “one fix” |
Kling 3 Turbo is 22 credits, discounted from 27, silent, 3–15s. Veo 3.1 is 64 credits, native audio, 4 / 6 / 8s. Seedance 2.0 is 44 credits, 4–15s, 1080p, audio on. Seedance 2.5 is 34 credits (was 67), 4–30s, 720p, audio on. Gemini Omni Flash Edit is 50 credits, edit-video. Read the card at confirm. Cards move. A spreadsheet that hard-codes last month’s rate is how a “cheap” week becomes a surprise.
Retries are the line producers forget. Five eight-second tries are forty seconds. If the row is per-second, the miss is not free. Put a retry cap on the hero (two, not ten). Put a cheaper silent row on blocking tests so the hero budget is not eaten by “seeing if the camera works.”
Captions are a separate pile. 21 basic presets at 1×, 9 dynamic at 2×, per clip. Thirty clips on Glass is sixty base caption renders. Budget the multiplier as a choice, not as a leftover.
Do not divide a still’s credits by a fictional duration to make a “per-second” comparison with video. Different billing types are not one number. A Nano Banana still is a still.
Why USD-per-second wrecks the Monday plan
Host posts will say Seedance is $0.067/s at 480p or some other cell this month (fal, Atlas Cloud, and similar pages have published vendor figures). Even when those figures are real for that host, they are not this catalog, they omit your retry rate, they omit captions, they omit the silent-versus-sounded toggle, and they omit that you cannot buy 7 seconds on a row that only sells 4 / 6 / 8.
A producer who budgets “ten seconds of Kling” in a host’s dollars has not named a slug. Kling 3 Turbo is not Kling O3 Pro. One is 22 credits and silent. One is 56 credits and sounded. Ten seconds of the wrong row is a different week.
A producer who converts credits into dollars with a homemade exchange rate is inventing a rate they do not actually know and will argue with finance when the pack changes. Quote credits to the team. Quote whatever your account shows as the pack price to the client, as a project fee, not as a per-second unit copied from a host.
Audio is a rate toggle on some rows and a hard off on others. Turbo is off. Veo can be sounded at a different per-second cell than silent. Budgeting “Veo” without the audio flag is not a number.
Duration ladders waste seconds you did not want. You cannot buy 7 on a 4 / 6 / 8 row. You buy 8 and trim, or you pick a row that exposes the seventh second. The unused second is still billed if the meter is per-second and the ladder forced you up. Put the ladder in the plan: “8s Veo, trim to 6,” not “6s of Veo.”
The Monday habit
- List clips you actually owe, not clips you might make if a generate is pretty.
- Assign a named slug per shot type. Hero, draft, pack, caption tier.
- Write tries: 1 lock + N retries, N small.
- Sum credits. Compare to the balance. Cut scope or cut retries before you generate, not after the pool is empty.
- Ban USD/s in the Slack thread. If someone pastes a host blog, ask which slug on this catalog they mean.
Mid-week, do not swap a dear row for a cheap row on a locked hero to “save the budget.” That is a new identity, which is a reshoot, which is not a saving. Swap rows on drafts. Protect the hero slug.
If the week is caption-heavy, decide 1× versus 2× on Monday. Dynamic as a Thursday panic is how the caption line doubles after the video line is already spent.
If the week is research — try every model on one prompt — budget it as a comparison run with a cap, not as production. Comparison is a cost scenario of its own. It is not “the first clip.”
Keep a running actuals column: credits quoted at confirm, not credits you remembered. The app quotes the whole-credit total. Trust that number over the plan if they disagree, then ask why (retry, audio on, duration stepped up).
What the habit is for
So that Thursday you still have credits for the hero, not only for the eight Turbo tests you ran because they felt free.
So that a client who forwards a host’s $ per second does not set your price.
So that Seedance 2.5’s 34 and Veo’s 64 are compared as cards plus ladders plus retries, not as one headline number each.
The cost hub is the worked arithmetic when you need a formula. The radio this week is: plan in credits on named rows, then generate.
If you need a balance that actually meters those rows, sign up and budget the slate in credits before the first confirm. USD per second is someone else’s catalog.
FAQ
Can I show the client a USD number at all?
You can quote a project fee from your pack and your scope. Do not present a host’s per-second API price as this studio’s unit cost, and do not invent an exchange rate from credits to dollars. Credits are the unit the confirm screen uses.
Why not budget in seconds, since so many rows are per-second?
Seconds without a slug are not a price. Ten seconds of Turbo is not ten seconds of O3 Pro. Budget seconds on a named row, which is a credit number. The second is an input. The credit is the budget.
The card changed mid-week. What do I do?
Read the new quote at confirm. Update the actuals column. Do not keep generating against a stale spreadsheet cell. If the new quote blows the week, cut retries or cut clips, not the hero slug.