CREDITS & PRICING • Framework updated · Jul 2026
Which AI video plan fits your monthly clip target?
Choosing an AI video plan is really a math problem: how many clips you need each month, times the credits each clip burns, weighed against the fixed credit pool a tier hands you. Get that division right and you avoid both overpaying for a tier you never fill and stalling mid-month on a plan that runs dry. This guide turns a monthly clip target into the minimum plan tier, then shows the break points where upgrading beats buying top-ups.
Start with a target you can actually compute
Define your monthly target two ways: number of clips and total finished seconds. Duration and resolution can push you past a tier even when the clip count looks small, so both numbers matter. Then find the credits each clip costs at your chosen model and settings. On Runway, the flagship Gen-4.5 model runs 60 credits per 5 seconds of video, which is 12 credits per second, per the official pricing table. From there the formula is simple: target clips multiplied by credits per clip equals your monthly credit demand. Treat that demand as the number every tier must clear. A worked version: if you plan T five-second clips, you need 60 times T credits before any retries. Then pad it for failed generations, because a clip you re-run twice costs the credits three times. That padded number, not your ideal count, is what actually sizes the plan you should buy.
Let the credit pool pick the minimum tier
Each paid tier is a fixed monthly credit pool, so the smallest pool that covers your padded demand is your floor. Runway's annual-billing ladder runs Standard at $12 per month for 625 credits, Pro at $28 for 2,250 credits, and Max at $76 for 9,500 credits. The jumps are large and uneven: roughly 3.6 times from Standard to Pro, then about 4.2 times from Pro to Max. At 60 credits per five-second clip, Standard's 625-credit pool covers about ten clips before it is spent, Pro clears several times that, and Max is built for heavy daily output. Map your padded demand onto those pools and the tier chooses itself. If demand lands just above a tier's ceiling, flag it. That narrow overshoot is exactly the situation where a one-off top-up, not a full upgrade, is usually the cheaper fix.
Free tiers rarely count toward a recurring target
A recurring monthly target needs a recurring monthly pool, and most free tiers do not provide one. Runway's free plan is a one-time 125-credit trial with three video editor projects and 5GB of storage, not a monthly refill, and watermark-free output is listed as a paid feature starting at Standard. Luma's Dream Machine free plan is capped to draft resolution with watermarks and non-commercial use; its iOS free plan lists 250 monthly credits under those same limits. Pika's free plan is limited to the Pika 2.5 model at 480p and cannot remove the watermark, buy extra credits, or license clips commercially. Some providers go further and structure free credits as small daily grants that expire the same day, which cannot be banked toward a monthly goal, so check how long free credits last before counting them. For anything you intend to publish or monetize, start at the first paid tier and treat free credits as evaluation only.
The expire-versus-persist line is where upgrades beat top-ups
The break point between upgrading and buying top-ups sits on one boundary: subscription credits usually expire, while purchased credits usually persist. Runway spells this out plainly. On Standard and Pro, monthly credits reset within 24 hours of your billing date and do not roll over, whereas separately purchased credits never expire. Several other providers follow the same shape, with purchased top-ups carrying over long after the monthly grant resets. The practical rule falls out of that: size your base tier to the demand you hit almost every month, and use persistent top-ups to absorb occasional spikes. Buying top-ups every single month is a signal you have outgrown the tier and an upgrade is now cheaper per clip. Buying them once a quarter for a busy stretch is exactly what they are for. Match the tool to the pattern: steady demand to the tier, spikes to top-ups.
Top tiers change the math with rollover and unlimited modes
At the top of each ladder, the structure itself can flip the decision. Runway's Max tier lets monthly credits roll over for one month, so buyers who undershoot some months stop forfeiting value the way they would on Standard or Pro. Luma's Dream Machine Unlimited plan, at $94.99 per month, pairs 10,000 fast-mode credits with unlimited generations in Relaxed Mode, and that Relaxed Mode is exclusive to the Unlimited and Enterprise plans. Both carry the same lesson: once your target is high and steady, the top tier can remove the per-clip ceiling or preserve unused value in a way that endless top-ups on a mid tier never do. The break point is sustained volume. If you are topping up a mid tier month after month and still running dry, price the top tier's rollover or unlimited mode against that recurring top-up spend before assuming the cheaper sticker is actually cheaper.
Confirm the product line, resolution, and current numbers
Two traps can invalidate an otherwise clean calculation. First, resolution and duration can force a higher tier than clip count alone. Pika's free tier is capped at 480p, while its paid Standard, Pro, and Fancy tiers unlock all resolutions, and duration and feature options vary by tier across providers. A target defined in finished seconds, not just clip count, can jump you a tier. Second, confirm which product you are pricing. Luma publishes two different official ladders: the Dream Machine plans documented in its support hub, and a separate subscription ladder on its main pricing page whose Plus, Pro, and Ultra tiers list 10,000, 40,000, and 150,000 credits at $30, $90, and $300 per month. They are not the same product. Prices and pools also change often, so verify every figure on the provider's live pricing page before you commit, especially any tier where the annual and monthly options differ.
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.