On this page
AI Video Costs: Credits, Subscriptions and Retry Budgets

At the settings checked below, a five-second video costs 60–162 credits. That is the price of one completed attempt, not a guarantee of a usable clip. Duration, model variant and resolution change the quote.
How many credits does one video cost?
Checked November 15, 2026, at 16:41 UTC, without generating videos. These are final Photo to Video quotes for image-to-video with sound, Spicy off and Turbo off. The named presets include sound; these are not quotes for silent variants. Resolution is the selected setting, not a guarantee of exact output dimensions.
| Model preset | Duration / resolution | One attempt | Two attempts |
|---|---|---|---|
| Wan Pro | 5 s / 720p | 60 credits | 120 credits |
| Wan Pro | 10 s / 720p | 120 credits | 240 credits |
| Seedance 2 Mini | 5 s / 720p | 72 credits | 144 credits |
| Seedance 2 Mini | 10 s / 720p | 144 credits | 288 credits |
| Kling Fast | 5 s / 720p | 84 credits | 168 credits |
| Kling Fast | 10 s / 720p | 168 credits | 336 credits |
| Seedance 2 Fast | 5 s / 480p | 84 credits | 168 credits |
| Seedance 2 Pro | 5 s / 480p | 162 credits | 324 credits |
Two attempts means paying for another completed generation if you want to try again. It does not mean two attempts will produce a result you keep. The 480p rows are different settings from the 720p rows, not a matched-quality comparison. Check Make Video for the current quote before submitting.
What does that mean in dollars? Using the publicly listed $4.99 / 150-credit pack checked on November 12, a 60-credit attempt consumes about $2.00 worth of that pack (60 ÷ 150 × $4.99), before any applicable tax. That is an allocation of purchased credits, not a $2.00 checkout option. The purchase is $4.99; larger packs have a different cost per credit. See the cash purchase examples below for actual pack combinations.
Not sure which preset to try? Use our model selection table with original test results. We keep the November 7 test charges below separate from this newer quote check.
What one five-second generation cost in our test
We used the same source portrait and prompt, selected 5 seconds and 720p, and generated two consecutive runs with each preset. We did not replace weaker results or select only a winning take. All six jobs completed; completion does not mean that every result met the same creative standard.
| Preset | Credits per run | Two runs |
|---|---|---|
| Wan Pro (Wan 3.0) | 60 | 120 |
| Seedance 2.0 Mini | 72 | 144 |
| Kling Fast | 84 | 168 |
Total: 120 + 144 + 168 = 432 credits. The complete outputs, source, prompt and individual records are in our Kling vs Wan vs Seedance portrait comparison. The selected settings did not produce identical dimensions or exact five-second files; see that comparison for the actual exports.
These are November 7 charges. A read-only check on November 8 returned the same three button quotes, without generating additional videos. Prices can change with the preset, settings and date. Check the quote on Make Video for your own setup.
Budget for 10 kept clips
Generation budget = credits per completed attempt × assumed completed attempts per kept clip × number of kept clips. When settings differ between attempts, add the individual quotes instead of using one rate for everything.
The table below assumes ten kept clips and the same selected 5-second, 720p preset throughout. The attempt counts are illustrations only. We have not generated ten usable clips for this guide or measured a success rate.
| Preset | 1 attempt each | 2 attempts each | 3 attempts each |
|---|---|---|---|
| Wan Pro (Wan 3.0) | 600 | 1,200 | 1,800 |
| Seedance 2.0 Mini | 720 | 1,440 | 2,160 |
| Kling Fast | 840 | 1,680 | 2,520 |
All numbers are credits, not dollars. One attempt assumes every first result is kept. Two and three attempts allow for more completed results being discarded. Neither is a recommendation based on typical user behavior; you could need more than three, or never obtain the result you intended.
For example, ten Kling Fast clips at the recorded 84-credit rate would cost 840 credits if every first result were kept. Twenty completed attempts would cost 1,680 credits; thirty would cost 2,520. Those attempt counts do not guarantee ten keepers.
Check the quote for your video
What would 3, 10 or 30 kept clips cost to buy?
Prices checked November 12, 2026, in the signed-in web purchase panel: 150 credits for $4.99; 500 for $12.99; 1,200 for $24.99; 2,500 for $49.99, in USD. A separate read-only button check returned 60 credits for Wan Pro, 5 seconds, 720p. No purchase or new generation was made for this price check.
Assume zero starting balance, no welcome credits, two completed attempts per kept clip, and that every attempt uses this 60-credit preset. The purchases below are the lowest listed pack combinations covering each budget, before any applicable tax. They are upfront cash outlays, not a measured cost per successful result.
| Kept clips planned | Attempts / credits | Purchase | Cash / credits left |
|---|---|---|---|
| 3 | 6 / 360 | One 500-credit pack | $12.99 / 140 |
| 10 | 20 / 1,200 | One 1,200-credit pack | $24.99 / 0 |
| 30 | 60 / 3,600 | Three 1,200-credit packs | $74.97 / 0 |
For 30 clips, three 1,200-credit packs cost $74.97. A 2,500 plus a 1,200 pack costs $74.98 and leaves 100 credits instead. Buying the largest pack is not automatically the smallest checkout bill. Subtract your existing balance before choosing packs; do not repeat the zero-balance purchase calculation each month if you still have credits.
With a $20 spending limit: a 500 plus a 150 pack costs $17.98 before tax and supplies 650 credits: ten completed attempts at 60 each, with 50 left. Under the two-attempt assumption, that budgets for five kept clips, not ten guaranteed keepers. Cheaper presets change the arithmetic; a pack’s advertised maximum video count is not the capacity at every setting.
Credit packs or a monthly subscription?
Photo to Video’s web Terms describe one-time, non-renewing credit packages. Our November 12 system check found no monthly reset of unused purchased credits. The Terms do not promise permanent validity, so treat this as the current behavior rather than a lifetime guarantee. Free and promotional allowances have separate conditions.
For a subscription example, Runway Standard lists $15 on monthly billing with 625 credits per month; the $12 monthly equivalent requires annual billing ($144 for a year). Its Gen-4.5 documentation lists 12 credits per second and 720p output, making one five-second generation 60 Runway credits. Checked November 12, 2026.
Here are illustrative Runway purchases for the same 6, 20 or 60 attempts, using Standard plus additional credits where needed. Additional credits start at $10 for 1,000. These examples exclude free credits, tax and other features; they are not an exhaustive optimization of every plan or top-up amount.
| Kept clips planned | Gen-4.5 attempts | Illustrative cash purchase | Unused balance |
|---|---|---|---|
| 3 | 6 | $15 Standard | 265 monthly credits |
| 10 | 20 | $15 Standard + $10 top-up = $25 | 425 purchased credits |
| 30 | 60 | $15 Standard + $30 top-up = $45 | 25 purchased credits |
These models are different. Matching five seconds and a 720p setting does not match image quality, sound, motion accuracy, exact export dimensions or the number of attempts you will need. We did not generate Runway comparison footage for this table. Both happen to quote 60 of their own credits here; credits are not interchangeable across services.
On Standard, unused monthly credits expire at the next billing cycle; separately purchased credits do not expire and monthly credits are spent first. Runway subscriptions renew until cancelled; annual billing still allocates credits monthly. After cancellation, purchased credits remain, but available models and features depend on the resulting plan.
Choose from your workload: occasional work favors avoiding a recurring bill you may forget or an allowance you will not use. Regular work can favor a subscription—even the illustrative 30-clip Runway purchase above costs less cash than our selected Wan pack scenario. First test whether the specific model meets your needs, then compare actual spending across active and inactive months. Neither “no subscription” nor a large credit allowance proves better value.
Make a budget for your actual project
- Define what you will keep. Decide whether recognizable facial detail, a particular hand movement, framing or duration is essential. A completed file can still fail that creative check.
- Record the exact setup and quote. Note the model variant, duration, resolution and any other available settings. A different preset is a different budget.
- Run a small test before a batch. Review the whole output. Our prompt guide can help you describe the intended action clearly; it cannot guarantee fewer attempts.
- Set a spending limit. Choose how many additional completed attempts you are willing to pay for. Stop and reconsider the source, action or model when you reach that limit.
- Track actual charges and kept clips. Divide the net credits spent on the project by the number you kept to learn your own cost per kept clip. If you kept none, report the spend and zero keepers rather than inventing a rate.
In the portrait comparison, Kling Fast retained more of the starting expression and hair across the face; Wan made the smile more pronounced, and Seedance changed facial detail and framing more. That is evidence about one source and prompt, not a general identity guarantee. Read the full six-run comparison before choosing solely by the price.
A redo, a failed job and an extra editing step are different costs
A completed video you choose not to use still consumes credits. Generating another version is another attempt. Do not subtract discarded videos from your spending record.
Failed paid generations normally return charged credits automatically, but Grok Imagine content-policy failures do not. A pending job is not a failed job. Check your balance and transaction history for the actual credit return; contact support if a failed job’s eligible return is missing.
Separate test-account records checked on November 11 show the difference:
- Submission rejected: two Kling requests on November 11 each deducted 84 credits and returned 84 in the same second. The net cost was zero, but both a charge and a return occurred.
- Generation failed: a Kling Pro task on November 9 deducted and returned 126 credits; a MiniMax H3 Turbo task deducted and returned 144. Each charge and return was linked to the same failed task.
- Completed, but unwanted: our November 7 first jacket replacement attempt left the clothing unchanged and cost 45 credits, with no credit return. The second attempt cost another 45.
These are recorded examples, not a guaranteed return time. Returning credits to your balance is different from refunding a credit-pack purchase; cash-refund conditions are in our Terms of Service. Clearing an in-progress task in the App does not stop generation or trigger a credit return.
The budget table excludes optional upscaling, video editing and extensions. Add their current quotes separately if your project needs them. A higher-resolution export does not by itself prove the result is usable.
Eligible new accounts may receive welcome credits, subject to eligibility limits. A 50-credit balance is insufficient for any of the three 60-, 72- or 84-credit settings here. Check your actual balance and a lower-priced configuration, or buy a credit pack if needed. See the free-credit guide for the conditions.
The cash scenarios above use dated purchase prices. A balance you already own reduces the next purchase you need; allocating a cash value to credits consumed is different from the actual amount charged at checkout.
What this budget cannot tell you
- The six-run ledger is measured; the ten-clip budgets are hypothetical.
- Two runs per preset cannot establish a success rate, an average retry count or a universal best model.
- The sample covers one portrait, one prompt and the selected 5-second/720p settings. It does not price every workflow on Photo to Video.
- The cheapest individual generation is not necessarily the cheapest route to your creative goal. Compare actual results and total spend.
Start with the workflow in our photo-to-video guide, then use your live quote and a spending limit you are comfortable with.
Check Wan settings and credits
FAQ
Should I buy credits or a subscription for occasional AI videos?
Compare the cash needed for your exact settings and expected attempts, plus renewal and unused-credit rules. Photo to Video sells one-time credit packs. A subscription can be cheaper for some workloads, but unused monthly credits may expire. Neither billing model guarantees usable results.
Does Photo to Video renew automatically or reset credits monthly?
Web credit packs are one-time purchases, not subscriptions, so they do not renew automatically. As checked on November 12, 2026, the current system does not reset unused purchased credits monthly. Our Terms do not promise permanent credit validity; check the purchase terms before buying far ahead.
What can a $20 budget buy at these settings?
At the November 12 web prices, one 500-credit pack plus one 150-credit pack costs $17.98 and provides 650 credits. At 60 credits per Wan Pro 5-second 720p generation, that covers ten completed attempts with 50 credits left. It does not guarantee ten usable clips; taxes, if applicable, are additional.
How many credits should I budget for 10 AI videos?
Multiply one generation’s quote by assumed attempts per kept clip, then by 10. At 60 credits and two attempts each, that is 1,200 credits—a planning scenario, not ten guaranteed usable videos.
Do I pay for completed videos I do not keep?
Yes. Completed generations use credits even if you discard them. Each new attempt is charged separately.
Do failed generations return credits?
Failed paid generations normally return charged credits automatically. Grok Imagine failures caused by content-policy checks are an exception. A pending job is not a failed job; check your balance and transaction history.
Are two attempts normally enough?
We do not know. The six-run test did not measure attempts needed for an acceptable result; the one-, two- and three-attempt budgets are assumptions.
Which tested preset cost the fewest credits?
Wan Pro at 60 credits, followed by Seedance Mini at 72 and Kling Fast at 84, at our November 7 test settings. This is not a ranking of all models.
Can 50 welcome credits pay for these settings?
No: all three tested settings cost more than 50 credits. Choose settings within your balance or add credits.
Does this budget include upscaling or editing?
No. The table covers image-to-video generation only; budget separately for upscaling, editing and extending, using each action’s quote.
For a longer attempt with a refunded failure, see our 15-second timed-prompt test, with full outputs and the actual charges.