Veo vs Kling vs Runway: What One Usable Ad Clip Costs

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Veo vs Kling vs Runway: What One Usable Ad Clip Costs.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Veo vs Kling vs Runway: What One Usable Ad Clip Costs.

Google's Gemini Omni, which replaces Veo's preview models, is cheapest per clip on list prices: about $0.07 a second at 720p with audio. Runway's Gen-4.5 is about $0.75 to $1.15 for a five-second clip, and Kling 4.0 costs about $1.47 to $2.27 through Runway. Retakes matter more than the rate card.

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

The title says Veo, but the premise has moved. Google's own deprecation page lists the Veo 3.1 preview models with an earliest shutdown of 22 October 2026 and points to Gemini Omni 1.2 Flash, released on 5 November 2026. This tutorial therefore prices the model Google now recommends, and keeps Veo's older per-second figures in a clearly marked box. Every price here is a list price in dollars; treat it as a starting point and check the vendor's page on the day you buy.

What should a cost per finished ad clip include?

A cost per finished clip has four parts, because a rate per second is not what you pay for an ad. The number you want is the cost of one clip you would actually publish.

  1. The generation itself: seconds times the rate, in credits or dollars.
  2. Retakes: the generations you throw away. If you keep one in three, multiply by three.
  3. Extras: sound, upscaling, edits and any second tool you need to fix a clip.
  4. Your time: writing the brief, reviewing every attempt, and assembling clips into an ad.

The first part is public. The second is yours to measure. I have not run the same brief through all three tools, so the retake rates below are assumptions for the sums, not results. The scorecard near the end lets you measure them on your own brief for a few dollars.

How does Runway price a clip in credits and dollars?

Runway prices clips in credits per second, and its pricing page, the most transparent of the three, ties plans to credits and credits to seconds. These are the annual-billing figures shown on the page this month.

PlanAnnual price a monthCredits a monthDollars per credit
Free$0125, one-timen/a
Standard$12 ($15 billed monthly)625$0.0192 ($0.024 monthly)
Pro$282,250$0.0124
Max$769,500$0.0080

Runway's FAQ says Gen-4.5 uses 12 credits per second of generated video, so a five-second clip is 60 credits and a ten-second clip is 120. Gen-4 Turbo is cheaper per second, about 5 to 6 credits, but the page's own figures disagree slightly between plans, so check before relying on it. Kling 4.0 is also offered inside Runway at 118 credits per five seconds at 1080p, with a 25% discount that takes it to 89 until 30 November 2026.

Five-second clip on RunwayCreditsStandard annualPro annualMax annual
Gen-4.560$1.15$0.75$0.48
Kling 4.0, 1080p118$2.27$1.47$0.94
Kling 4.0, 1080p, to 30 Nov promo89$1.71$1.11$0.71

Two points. First, the price per credit falls sharply as you move up the plans, so the same clip costs about 40% less on Pro than on Standard. Second, Standard and Pro credits do not roll over, so unused credits at month end are money lost. Runway's FAQ says top-up credits can be bought with a 1,000-credit minimum and never expire; the page does not give their price here, so ask before you count on them. For Runway's plan-by-plan maths in more depth, see is Runway worth it for video ads: Standard vs Pro credit maths.

How much does a Kling 4.0 clip cost?

Kling publishes no per-clip cost for Kling 4.0 in its own app: it gives the value of a credit but not the credits a 4.0 clip uses. Its credits policy says $1 buys 66 purchased credits, purchased credits are valid for two years, regular membership credits for one month, and failed generations are refunded. Kling's membership prices and the credits each plan includes change often and are not quoted here, so read them on Kling's own membership page before you buy.

That gives you a credit price: about $0.015 at the top-up rate ($1 divided by 66) or about $0.013 on the membership figure ($8.80 divided by 660). What it does not give you is how many credits a Kling 4.0 clip uses in Kling's own app. Kling's pages state no per-clip figure for it, and Kling's API pricing page says rates vary by model, mode, resolution and whether audio is included, with the table itself not readable for me. So the method is: make one test clip, note the credits deducted on your account, and multiply by the dollars per credit.

The only Kling 4.0 price I could verify per clip is the one inside Runway above. If you use Kling directly, your cost depends on your plan and mode, so measure it. The site's own tutorial on what a usable Kling clip costs in credits goes through that measurement.

What does Google's Veo successor cost per second?

Gemini Omni 1.2 Flash, Google's replacement for the Veo preview models, costs about $0.07 a second at 720p. Google's pricing page for the Gemini API lists it as a video generation and editing model producing up to 15 seconds with native audio, on the paid tier only. Output is billed in tokens, which Google converts at 5,792 tokens per second of 720p video.

Gemini Omni 1.2 FlashApprox. per second5 seconds10 seconds15 seconds
720p$0.07$0.35$0.70$1.04
1080p with upsampling$0.104$0.52$1.04$1.56
360p (drafts)$0.023$0.12$0.23$0.35

Google's page quotes the 10- and 15-second 720p figures directly; the others are my multiplication of its per-second rates. The previous model, Omni 1.1 Flash, is deprecated and from 3 December 2026 its requests are routed to 1.2 Flash at 1.2 rates.

The cheap drafts are worth noticing: at 360p you can test ten ideas for about $2.30 before paying for a 1080p render of the one that works. That changes how you work, because experimenting is almost free.

Veo's older figures, with a warning

Google's deprecations page lists the Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite preview models with a shutdown date of 22 October 2026 and names gemini-omni-1.2-flash as the recommended replacement, so this tutorial does not quote Veo's per-second rates. Do not plan around the preview models; the Omni rates above are the ones on Google's pricing page at the time of writing. If you reach Veo through a Google plan or app instead of the API, read Google Veo pricing: what AI Plus, Pro and Ultra credits buy, and note that the Gemini app's video now runs on Omni rather than Veo.

How do Veo's successor, Kling and Runway compare on one page?

On list prices, Gemini Omni 1.2 Flash at 720p is the cheapest five-second attempt in the table at $0.35, but retakes matter more than the rate card.

This table puts all three on the same footing: a five-second clip at a sensible resolution, on the cheapest sensible plan, with the unknowns left blank on purpose.

ToolCost of one 5-second attemptAudioCredit expiryWhat is unknown
Gemini Omni 1.2 Flash, 720p$0.35 (pay as you go)Native audio per Google's pageNo credits; billed per usePaid tier needed; rights terms to check
Runway Gen-4.5 on Pro$0.75Not verifiedPlan credits do not roll overCommercial terms and training use to check
Kling 4.0 inside Runway on Pro$1.47 ($1.11 to 30 Nov)Not verifiedPlan credits do not roll overDirect Kling cost per clip unpublished
Kling in its own appMeasure it: credits x about $0.013 to $0.015Varies by mode per Kling's API pageMembership credits one month; purchased two yearsCredits per clip

On rate alone Google is cheapest, then Runway's own model, then Kling inside Runway. But rate is the smallest part of the story once retakes enter.

How many retakes decide which tool is cheapest?

Retakes decide which tool is cheapest in practice. Take a four-person online shop that wants a five-second product clip for a social ad: a ceramic mug on a wooden table, steam rising, soft morning light. The owner needs ten usable clips for the month. These retake rates are assumptions to show the effect, and I have not measured them.

Attempts per usable clipOmni 720p ($0.35)Runway Gen-4.5 Pro ($0.75)Kling 4.0 in Runway Pro ($1.47)
2$0.70$1.50$2.94
3$1.05$2.25$4.41
5$1.75$3.75$7.35
8$2.80$6.00$11.76

Look at the diagonal. A tool at $0.75 that gets a usable clip in two attempts ($1.50) beats a tool at $0.35 that needs five ($1.75). So if one tool understands your product shots and the cheaper one does not, the dearer rate card wins. For ten clips, the gap between the best and worst cells in a row is only a few dollars to a few tens of dollars; the real cost is your time reviewing fifty attempts.

The corollary is that a plan's monthly allowance matters. Pro's 2,250 credits buy 37 five-second Gen-4.5 attempts, or 19 Kling 4.0 attempts at 118 credits. Ten usable clips at three attempts each is 30 attempts, which fits on Pro for Gen-4.5 (1,800 credits) but not for Kling 4.0 (3,540 credits), which pushes you to Max or a top-up. That is a cost the per-clip table does not show.

If you keep the retake sums in a spreadsheet, try The Complete Excel Course with AI Integration on Coding Liquids.

How do you test one brief on all three tools?

Test one brief on all three tools with about $15, because you should not trust anyone's retake rate, mine included. Use the free credits and low-resolution drafts first.

BRIEF (same words in every tool)
"Five seconds. A ceramic mug on a wooden table by a window, morning light,
steam rising, camera slowly pushes in. No text, no people."

Tool:            Attempts   Kept   Credits/$ spent   Seconds of my time   Notes
Omni 720p           [ ]      [ ]       [ ]               [ ]
Runway Gen-4.5      [ ]      [ ]       [ ]               [ ]
Kling 4.0           [ ]      [ ]       [ ]               [ ]

Usable = I would publish it unedited:  yes / no
Flaws to log: warped handle, wrong steam, text artefacts, flicker, drift

Run five attempts per tool, without changing the brief between them. Count a "kept" clip only if you would publish it. Divide spend by clips kept to get your cost per usable clip, then add your minutes at your own hourly value. An illustrative result might be Omni 5 attempts, 2 kept, $1.75, so $0.88 a usable clip. Your numbers will differ, and they are the ones that matter.

How do you cut retakes by changing the brief?

You cut retakes by changing the brief, not the tool, because every wasted attempt costs the full rate. AI video models tend to do best with one subject, one camera move and a plain background, and worst with text, hands, small moving parts and anything that must match a real product exactly. Here is an illustrative before and after for the mug clip.

BEFORE
"A lifestyle ad for our new mug with someone enjoying coffee, our logo on
the side, cosy kitchen, steam, they smile, text saying 'Mornings made better'."

Likely faults: warped logo, odd hands, garbled on-screen text, a face that
drifts between frames. Typical outcome: many attempts, few usable.

AFTER
"Close-up of a plain white ceramic mug on a wooden table by a window. Soft
morning light. Steam rises slowly. The camera pushes in very slightly.
No people, no text, no logo."

The after brief drops everything the model is likely to get wrong. You add the logo and the headline in an editor, where they stay sharp and exactly on brand, which also means the AI part of the clip carries no brand risk. If your retake rate is five attempts for the before brief and two for the after brief, the saving on Runway Gen-4.5 is 3 x $0.75, or $2.25 a clip, plus the time you no longer spend reviewing bad attempts. Over twenty clips that is $45 and a good hour or two of your time.

A second example: product clips from a still photo

Some tools animate a photo you supply, which can be the surest route for a real product because the model starts from your exact item. If your tool of choice supports it, a five-second move on a product photo with a gentle push-in is a lower-risk first test than generating the product from words. Check each tool's current image-to-video options and costs on its own page, because I have not verified them here, and run the same scorecard on them.

What does the rate card leave out?

A rate card leaves out sound, upscaling and edits, which all raise the price of a finished clip.

Sound

Google's page describes Omni as generating with native audio. Kling's API page says its prices vary by whether audio is included. Runway's pricing page does not state whether Gen-4.5 clips include audio, so add the cost and time of a separate music or voice step if you need one. An ad with no sound is fine for muted autoplay but rarely for anything else.

Length and resolution

Most social ads are five to fifteen seconds. Omni's 15-second ceiling suits a single scene; Runway's cost scales linearly with seconds, so a ten-second clip is twice the credits. If you need a thirty-second ad, plan for several clips stitched in an editor, and for continuity between them, which is where retakes multiply.

Rights and data

Ads have more risk than experiments. A previous check recorded that Runway allows commercial use but may train on uploads and outputs. Kling's credits policy is silent on commercial use, so read its terms and the site's note on Kling's data, watermarks and rights. Google's terms for Omni apply to anything you make with it. Keep a dated copy of the terms for every tool you use in paid advertising, and never upload a customer's photo or a product under embargo without checking.

Platform rules

Ad platforms have their own rules about AI-generated content and disclosure. Check the platform's current policy before you spend on media, as covered in Google Veo for ads: make a short video ad and check platform rules.

A realistic first month, priced

An architecture practice wants six short clips for its project pages: a time-lapse of a building elevation, a walk-through of a courtyard, three material close-ups and a logo sting. Suppose it spends the first week on drafts and then renders finals. Using Omni for drafts at 360p ($0.12 for five seconds) and finals at 1080p ($0.52), with an assumed four draft attempts and two final attempts per clip, one clip costs about 4 x $0.12 + 2 x $0.52 = $1.52, so six clips cost about $9. The same six clips on Runway Pro at an assumed three Gen-4.5 attempts each is 6 x 3 x $0.75 = $13.50, but uses 1,080 of the 2,250 credits, so the plan fee of $28 is the real cost unless the rest of the allowance is used. In that case Runway's cost per clip is $28 divided by six, about $4.67, until the other credits find a use. The takeaway: pay-as-you-go costs less for small volumes, and plans cost less when you actually use the allowance.

Keep a running log so the next month is cheaper

After the first month, keep a one-line record for every clip you publish: tool, brief, attempts, spend and a note on what went wrong in the rejects. After thirty clips you will see patterns, such as a tool that always warps handles or a camera move that always drifts. Those patterns are worth more than any rate card, because they tell you which brief to send to which tool, and which shots to film instead.

How to decide, in order

  1. Count the clips you need each month. Under about ten, pay-as-you-go pricing usually beats a plan.
  2. Run the scorecard with five attempts per tool on your hardest product shot.
  3. Compare cost per usable clip including your minutes, not the rate card.
  4. Check rights and platform rules for the tool that wins, and keep a copy of the terms.
  5. Choose a plan only if you will use most of its credits, and note that Standard and Pro credits do not roll over.

For the wider question of whether any of this beats filming, how much an AI marketing video costs compared with a shoot is the next read. If you would rather test a brief and a budget with someone, that is what my AI implementation consultation covers.

Questions about the cost of AI ad clips

Which is cheapest per usable ad clip, Veo, Kling or Runway?

On list prices, Google's Gemini Omni 1.2 Flash is cheapest, at about $0.35 for a five-second 720p attempt against about $0.75 for Runway's Gen-4.5 on Pro. Kling publishes no per-clip cost for Kling 4.0 in its own app. The real winner depends on retakes, so test one brief on all three before you commit.

Is Veo still the Google video model to price?

Not the Veo 3.1 preview models. Google's deprecation page lists them with a shutdown date of 22 October 2026 and names Gemini Omni 1.2 Flash as the replacement. Price Omni for the Gemini API and check Google's current pages for any Veo access you plan to use.

Why is cost per usable clip different from cost per second?

Because you rarely keep the first attempt. If one attempt in three is good enough to publish, your real cost is three times the price of a single generation, plus any editing. Retake rate often matters more than the rate card.

Does Kling refund failed generations?

Yes, if the generation fails. Kling's credits policy says credits are deducted when you generate and refunded if the generation fails. A clip that generates but looks wrong is not a failure, so it is not refunded; count it as spent and build that into your retake sum.

Do Runway credits expire?

Top-up credits do not expire, but most plan credits do. Runway's FAQ says credits bought as top-ups never expire, with a 1,000-credit minimum. Plan credits do not roll over on Standard and Pro, and Max rolls over up to a month of unused credits.

Can I use these clips in paid ads?

Each vendor's terms decide, and they change. Runway's help centre says you can use what you make commercially, and its Terms of Use let it train its models on your inputs and outputs. Kling's plan page lists commercial use as a paid-plan benefit, not a Basic one. Read the current terms and keep a copy before you run a paid campaign.

Further reads

Sources: Runway pricing page and FAQ; Kling credits policy and API pricing page; Google AI Studio pricing and deprecations pages. Veo per-second rates come from secondary reports and are marked as unverified.

Want a second pair of eyes on your setup?

Bring the process you're trying to fix. On a 1:1 call we'll work out which AI tools should handle it, which parts stay manual, and how to roll it out without disrupting the team.

Book a 1:1 call with me