CREDITS & PRICING • Framework updated · Jul 2026
AI Video Budget Template for Monthly Production
Paying for AI video is easy to underestimate because the sticker price of a plan rarely matches what a finished minute actually costs you. This guide gives you a monthly production budget template you can fill in yourself, with line items for generation, retries, editing, sound, review, and delivery. Each line uses documented assumption fields, so you plug in your own numbers and verify current rates on the provider's official page.
Start with your monthly credit pool
Every provider sells a monthly pool of credits per seat, so the first assumption field is which plan and how many credits it grants. On Runway, Standard includes 625 credits per month, Pro 2,250, and Max 9,500, while the Free plan gives 125 one-time credits. Luma's individual plans grant 10,000 credits on Plus, 40,000 on Pro, and 150,000 on Ultra. Pika lists 700 monthly video credits on Standard, 2,300 on Pro, and 6,000 on Fancy. Write your pool at the top of the sheet, then decide whether you budget per seat or, where supported, as a shared team pool; Luma's Team plan adds shared team credits and usage analytics rather than isolating credits to one login. Everything below draws down this pool, so treat it as your ceiling and verify the current number on the provider's pricing page before committing.
Line 1, generation: seconds times model rate
Generation is the one line the credit system prices directly, and the formula is clip seconds multiplied by your model's credits-per-second rate. Runway's API pricing shows how sharply the rate moves by model: Gen-4 Turbo runs 5 credits per second, Gen-4.5 12 credits per second, and Veo 3 40 credits per second. Resolution matters inside a single model too; Seedance 2.0 costs 36 credits per second at 480p or 720p, 40 at 1080p, and 150 at 4K. For a dollar view, Runway credits can be purchased for $0.01 each in the developer portal. Fill in three assumption fields per project: total finished seconds, the model you will use, and target resolution. Multiply them for every planned clip, sum the credits, and you have the raw generation line before any reruns. Because model rosters and rates change, confirm current per-second figures on the API pricing page.
Line 2, retries: a reroll and failure buffer
No first render is guaranteed to be usable, so the retry line separates budgeted output from real credit burn. Add an assumption field for your reroll rate: if you generate roughly two attempts for every clip you keep, your generation credits effectively double. Two provider details keep this line honest. Some models carry a per-generation minimum, such as Runway's Aleph 2.0 at 28 credits per second with a 56-credit minimum, which makes short reroll clips cost more than the raw math suggests. Working the other way, Luma automatically refunds credits for failed generations, so outright errors do not all count against your pool, though creative rerolls of successful outputs still do. Estimate your reroll multiplier from past projects, apply it to the generation line, and note whether your provider refunds failures. Verify both the minimums and the refund policy, since they change how much buffer you must reserve.
Lines 3 to 5: editing, sound, and human review
The credit meters cover generation only. Editing, sound, and human review are real costs that none of the three providers bundle into their credit pools, so they each get their own line and assumption fields. For editing, estimate hours of trimming, color, and assembly at your internal or freelance hourly rate. For sound, budget music licensing, voiceover, or sound design per finished minute, since generation credits buy no audio. For review, price the time a person spends selecting usable takes and checking brand and factual accuracy; on a monthly cadence this is often the most underestimated line. Enter each as hours times a rate, or a flat per-video figure if that is how you contract it. These three lines frequently exceed the generation credits themselves, which is why a template that stops at credits understates true monthly cost. Keep the rates you use documented so next month's estimate starts from evidence, not memory.
Line 6, delivery: tier-bound finishing costs
Delivery turns an approved take into a shippable file, and the key point is that some finishing options are plan-tier decisions, not per-clip add-ons. Watermark removal is the clearest example. Runway lists watermark-free output and 4K upscaling as included features from its Standard tier up, so a clean high-resolution file is a plan choice rather than a per-clip charge. Pika limits its Free and Basic tiers to Pika 2.5 at 480p, while Standard, Pro, and Fancy unlock all resolutions, and watermark-free downloads are listed on its paid plans. Commercial-use rights are also tier-bound: Pika lists commercial use as a feature of its paid tiers, and Luma includes commercial use in its paid plans. So your delivery assumption fields are the resolution you must ship, whether you need a clean watermark-free file, and whether the use is commercial. If any answer forces a higher tier, that cost belongs here, not hidden in the generation line. Confirm each tier's current terms before relying on them.
Reconcile against rollover and waste
The final step reconciles planned credits against how unused credits behave, because leftover credits are not always money saved. Rollover rules differ by provider and set your wasted-credit assumption field directly. On Runway, Standard and Pro credits do not roll over and reset within 24 hours of your billing date, while only Max carries up to one month of unused credits forward, and separately purchased credits never expire. Luma's monthly credits expire at the end of each billing cycle with no carryover. Pika treats rollover as a paid add-on you buy rather than automatic carryover. Practically, if your plan does not roll over, credits you do not spend each month are lost, so a pool far larger than your planned usage is overspending. Size the plan to your realistic monthly seconds plus your retry buffer, and revisit the tier whenever output volume changes.
Editorial note: This framework is general information, not a vendor endorsement. Check the current pricing, terms, and data-handling details directly with the provider before buying.