AI UGC ads cost: what 50 ad variants actually cost across Kling, Veo, and friends

Cost per variant, per provider

Planning length for this page: 15 seconds, audio on / default, one aspect, one prompt. Meta’s Facebook Feed video specs on 2026-09-18 accept 1 second to 241 minutes, so 15 seconds is inside that published window. Recommended Feed ratio is 4:5; the same tooltip says ratios from 16:9 to 9:16 are supported and may be masked to 4:5. Desktop Feed may show 1:1 with bars. TikTok’s official ads-spec HTML was not reachable when checked. This page does not treat 9:16 / 15s as a TikTok-official pair.

Twelve keys lotgen.ai pastes: Google, BytePlus ARK, fal, evolink, Kling, OpenAI, MiniMax, Black Forest Labs, Alibaba, ElevenLabs, xAI and Runway. Luma and Sora are models on fal or evolink, not keys. BytePlus ARK is the console that issues the Seedance key, and the name to put on the invoice.

15-second take, audio on where the key publishes a split. 1 / 3 / 5 attempts are a modeling assumption, not a measured keep rate. If the API will not bill 15 seconds, the row is clamped and labeled.

Closest official length when 15 seconds is not one take

How to read this table

• Dollars and seconds only. Cost per approved variant in this section is list × billed seconds × attempts at one aspect. Aspect re-renders and client rounds are the next section.
• 3 and 5 attempts are a model. There is no lotgen.ai observed keep rate on this page. Multiply by the takes you actually run.
• Veo 3.1 on the Gemini / Vertex predict path does not generate 15 seconds in one take. Veo 3.1 generate docs on 2026-09-18: video lengths 4, 6, 8 seconds; reference image-to-video only supports 8 seconds. An 8-second row is the honest Veo cell. A stitched two-clip model is in the 50-variant section, labeled as a model. The Veo-only walkthrough is Veo 3 cost per second.
• fal and evolink are their own checkout. Matching Kling’s 720p audio row on fal on the day checked is a coincidence, not a promise that 1080p or 4K match.
• BytePlus per-second is computed: official list 5.6 USD / M tokens for Seedance 2.0 Fast (input without video, 480p/720p), official formula (input seconds + output seconds) × width × height × frame rate / 1024, 24 fps, 1280×720, no input video, plus the one closing frame ARK bills (measured by lotgen on five real jobs; the page’s own examples reproduce without it, so these cells sit a cent or two above the page). 15s is 324,900 tokens, $1.82. The same page still prints a time-limited 25% off on 480p/720p Seedance 2.0 Fast through 2026-10-07 14:00 UTC+8 (about $0.09/s at 720p). This table uses the list rate. That window is dated copy, not a deadline on this page.
• Alibaba dollars are the International wan2.7-t2v row on Model Studio ($0.10/s 720P, $0.15/s 1080P). Official t2v duration is an integer from 2 to 15 seconds (default 5), so 15s is a billed length the API accepts. Do not paste evolink Wan as Alibaba list. The product catalog on live /pricing still prints Wan 2.7 1080p as unpriced; that is lotgen’s adapter catalog, not Alibaba’s page. 50 seconds on the pricing table is a free quota, not a duration cap.
• MiniMax dollars are official PAYG list. The video-package page still bills Hailuo 2.3 in points (768p/6s = 1 pt, 768p/10s = 2 pt, 1080p/6s = 2 pt). A 5% Standard package ($1,000 / 3,760 points) still infers about $0.28/point. That inference matches PAYG 768p 6s ($0.28) and 768p 10s ($0.56). It does not match PAYG 1080p 6s ($0.49). Use PAYG as list.
• xAI’s model pages print no cap; its video capabilities page says up to 15s, so the 15-second xAI cells are a length the API documents. Runway’s gen4.5 pricing row prints no cap either, and lotgen’s own test of the route (2026-09-06) tops out at 10 seconds, so Runway is a closest-length row, not a 15-second one.

1080p, audio on, 15s where the official page prints 1080p

Sibling compare (live): Kling vs Veo cost owns the 8-second Kling vs Veo dollar walkthrough. Sibling Kling credits (live): Kling AI pricing explained. The cost pillar across every key is how much AI video generation costs. This page does not replace it.

A 50-variant test: total cost by provider and attempt rate

This is a worked model, not a lotgen.ai run. No prompts were submitted. No takes were kept. The arithmetic is:

billed_seconds = clamp(requested_seconds, provider min, provider max)
cost_per_take      = list_rate × billed_seconds
cost_per_approved  = cost_per_take × attempts     # one aspect, before review
batch_total        = cost_per_approved × 50

Assumptions you can change: 50 approved variants, one prompt family each, 15-second billed length where the API will take 15, audio on / default, one aspect, attempts in {1, 3, 5}. Quality is not in the table.

Rounded to the nearest dollar in the 50-variant columns so a media plan can scan them. Exact cents used: Kling 720p audio 3× = $283.50; ARK 2.0 Fast 3× = $272.92; BFL hd 3× = $382.50; Alibaba 1080P 3× = $337.50; MiniMax 1080p 6s 3× = $73.50. Recompute from the one-take cells if you need invoice-level cents.

Worked Kling 3.0 API, 720p, native audio, 15s, 50 approved, 3 attempts

0.9 Units/s × $0.14 = $0.126/s
$0.126 × 15s = $1.89 per take
$1.89 × 3 attempts = $5.67 per approved variant
$5.67 × 50 = $283.50

Worked Veo 3.1 Fast 720p, 50 approved, 3 attempts, honest 8s cell

$0.10/s × 8s = $0.80 per take
$0.80 × 3 = $2.40 per approved 8s clip
$2.40 × 50 = $120

That $120 is not a 15-second variant budget. It is fifty approved 8-second clips.

Worked Alibaba Wan 2.7 t2v International 720P, 50 approved, 3 attempts, 15s

$0.10/s × 15s = $1.50 per take
$1.50 × 3 = $4.50 per approved variant
$4.50 × 50 = $225

Official duration is 2 to 15 seconds. This cell is a billed length the API accepts, not a stitch.

Stitched ~15s on Veo (model only). Two 8-second takes per attempt, then you cut. Not a lotgen run. Not a claim that the join will look like one shot.

Fast 720p:  $0.80 × 2 clips × 3 attempts × 50 = $240
Standard:   $3.20 × 2 clips × 3 attempts × 50 = $960

Edit time is extra and is not in these dollars. If the brief is “15-second Reels,” an 8-second-max model is a workflow tax, not a 15 × $/s line.

Sora 2 is a model on fal/evolink, not the OpenAI key. OpenAI’s own pricing page on 2026-09-18: Sora 2 $0.10/s at 720p, Sora 2 Pro $0.30/s at 720p, Pro 1080p $0.70/s. Batch is half. A 15s Sora 2 720p take is $1.50 if that marketplace row accepts 15 seconds. It does not belong on the OpenAI key.

Where the money actually goes

Retries are the first multiplier. They are not the only one. For a performance team or an agency, three other lines show up on the same job.

1. Retries before anyone reviews. A missed label, a bad hand, a talking-head that does not hold. The tables above already multiply by 1, 3, or 5. Change the multiplier. Do not cite an “industry keep rate.” Public provider pages checked for this page do not publish one.

2. Revisions after client or creative review. The keep is not always the ship. A common agency shape, labeled as a model: 40% of the 50 approved variants come back for 2 more takes.

Kling 3.0 1080p native audio, 15s:

review round = 0.40 × 50 × 2 takes × $2.52 = $100.80

That line is independent of the original 3-attempt generation. If your clients bounce half the board, use 0.50. If they bounce none, this line is $0.

3. Re-renders for aspect ratios. Meta Feed’s published recommendation on 2026-09-18 is 4:5 (1440 × 1800). The same Ads Guide tooltip says 16:9 to 9:16 are supported and may be masked to 4:5, and desktop Feed may display 1:1. A 9:16 keep that also has to run in Feed at 4:5 is a second render of the keep, not a second 3-attempt stack, unless the crop fails and you regenerate.

second aspect of the keep = 50 × $2.52 = $126    # Kling 1080p 15s, 1 extra take each

If you regenerate from scratch in the second aspect (new 3-attempt stack), that line becomes $378, same as the first aspect. Crop-then-check is the cheaper model. Full re-generation is the conservative one. Pick it in writing.

Worked “approved variant” stack, Kling 3.0 API 1080p native audio, 15s, 50 variants (model)

Same stack at Veo 3.1 Fast 720p, honest 8s (no stitch), one aspect, no client round: $120 for 50, $2.40 per approved 8s clip. Add a stitch (two 8s clips per attempt) and it becomes $240 before aspects and review. The brief asked for 15-second variants. The Fast cell only answers that brief if you accept 8 seconds or you budget the stitch.

Same stack at Alibaba Wan 2.7 t2v International 720P, 15s, one aspect, no client round: $225 for 50 at 3 attempts, $4.50 per approved variant. Add a second aspect of the keep: 50 × $1.50 = $75, total $300 before a client round.

BFL’s published table is useful here in a different way: draft t2v at hd is $0.06/s ($0.90 per 15s take) against $0.17/s full hd. Exploring variants at draft and paying full render on the keep is a documented BFL pattern, not a lotgen observation. Fifty draft takes + 50 full keeps at 15s hd: 50 × $0.90 + 50 × $2.55 = $172.50 before retries on the keep. If the keep still needs 3 full renders, you are back in the $7.65 cell.

Workflow that keeps cost visible

How to make UGC ads with AI, as a cost workflow rather than a prompt cookbook:

1. Brief the variant, not the second. Length the destination will actually run, aspect (4:5 Feed vs 9:16), audio on or off, the model and the tier (Veo Fast vs Standard is a 4× gap at 8s).
2. Quote the take in dollars on the keys you will paste, before the job runs.
3. Generate across more than one key when the brief allows it. Same prompt, two invoices, pick a keep.
4. Compare the takes with the quote still attached.
5. Share a review link so a client or brand manager signs off without a credit screenshot.
6. Approve, then spend the aspect re-render and the paid media.

lotgen.ai is built as that control plane: one place to generate across the twelve keys, see a real-dollar quote, compare takes, share a review link, and drive it via MCP. Invite-only beta. You still pay the provider. lotgen does not resell Kling credits or Veo tokens.

Request an invitationSeats and pricingRequest a team seat on the beta.

Brand / in-house teams can use the same form: request an invitation. Seat and plan detail: lotgen.ai/pricing. Public seat copy on 2026-09-18: $99 Solo, Team $90 a seat from 3, $79 from 6.

The how-to pillar, how to make UGC ads with AI, owns the production steps that are not cost. The media-mix side of variant testing is coming as a page of its own.

Billing clients for AI video

Agencies cannot invoice “some credits” and expect the client to reconcile it. The pass-through that holds up in a monthly recap is dollars, seconds, model, tier, attempts.

• Quote before generate. The line item is the same number the API will bill if the take succeeds. Kling developer API: failed video generations do not deduct. Gemini Veo: charged only if the video is successfully generated. MiniMax: failures or security-review videos do not deduct points. Alibaba Model Studio: failed requests incur no cost.
• Name the meter. Kling website membership is credits (owned by Kling AI pricing explained). Kling API is Units at $0.14. Mixing them is how “Kling is $X per video” disagreements start. MiniMax video packages expire in a month. Unused Kling API resource-pack balance does not carry over.
• Seat versus usage. lotgen.ai is a seat. Seconds still hit Google, BytePlus ARK, Kling, and the rest on the client’s or the agency’s keys, at list. Budget those two lines separately.
• Review is part of billing. A shared review link is how you prove which take was approved, at which quote, before the aspect re-render.

Do not bill an “AI vs human UGC creator” savings percentage unless you have the human invoice and you label the comparison. Vendor pages that advertise that gap are vendor claims. This page does not repeat them as fact.

The staffing and review-link process, in more depth, is coming as an agency workflow page of its own.

When AI UGC is not the right tool

Bulk AI video generation for ads is a good fit when you need many prompts, you can live with a keep rate, and the destination will take the length the API bills. It is the wrong tool, or only part of the tool, in these cases. Limits below are tied to public docs or labeled as product facts.

• The brief is 15 seconds and the model maxes at 8. Veo 3.1: 4, 6, 8 seconds. MiniMax Hailuo 2.3 at 1080p: 6 seconds only. Stitching is an edit, not a list price. Alibaba Wan 2.7 t2v is the opposite case: official duration is 2 to 15 seconds.
• A specific person has to be on camera. Likeness, talent, and usage rights are not a per-second row. Avatar UGC vendors rank for “ai ugc video generator”; lotgen.ai is not an avatar tool. This page does not price those products.
• On-pack legal copy, claims, or a logo must be pixel-true. A keep-rate model assumes you can throw a take. Packaging that cannot be wrong is a different QA bar. Provider docs do not publish a “text rendering SLA.”
• The client contracted a named model and tier. “Veo 3.1” without Fast / Standard / Lite is a 4× invoice gap at 8 seconds ($0.80 vs $3.20 per take at 720p). Get the tier in writing before you generate.
• Hands, product-in-hand, and lip sync. This page has no takes to show. Do not claim a winner. Same-prompt Kling vs Veo is the footage page, and it is not this one.
• You only hold a consumer credit plan and the job is API volume. A Kling membership does not fund a Kling developer key. See Kling AI pricing explained.
• US-only BytePlus ARK direct. Product fact from the calculator: US workspaces cannot use ARK direct. The Seedance row still belongs for keys that can paste it. US readers take fal/evolink Seedance, at marketplace rates.

Budget calculator

The calculator behind these pages is dated the same way: length, resolution, audio, usable shots (variants), attempts, the twelve keys.

The link below opens it with the agency preset: 15 seconds, 1080p, audio on, 50 shots, 3 attempts, the client toggle on. Pick the tile that fits, Agency or studio or Brand or in-house team, so the quote is worded for your side of the invoice.

The 3-attempt slider stays a model you set, not a measured keep rate. Veo rows will clamp to 8 billed seconds. Capture stays the live /request form. The calculator and this page read one dated rate table, so the Alibaba and MiniMax rows agree.

Until you open it, recompute from the tables above with the same formula as the 50-variant section.

billed_seconds = clamp(requested_seconds, provider min, provider max)
cost_per_take      = list_rate × billed_seconds
cost_per_approved  = cost_per_take × attempts
batch_total        = cost_per_approved × approved_variants

Prices checked on 2026-09-18: how to re-check yourself

Refresh this page quarterly, and within a few days of a major model or plan change.

1. Kling Units and per-second rows: kling.ai/dev/pricing. Confirm 1 Unit = $0.14 for video, then the Kling 3.0 720p/1080p/4K cells with and without native audio. Re-read duration on Kling’s own video API docs; this page used evolink’s 3-15s copy plus lotgen’s own test.
2. Veo 3.1 Gemini per-second rows: Gemini API pricing, section “Veo 3.1”. Confirm Standard / Fast / Lite. Google’s page was last modified 2026-09-16 when checked.
3. Veo 3.1 lengths: Veo 3.1 generate (4, 6, 8 seconds). Silent vs audio: Vertex generative AI pricing.
4. BytePlus ARK: ModelArk pricing. Confirm Seedance 2.0 Fast 5.6 USD/M tokens (list), whether the 25% off window is still printed, and 1080p support. Recompute tokens with the published formula. Do not paste a remembered 4.2 promo rate as list.
5. fal Kling: fal Kling Video v3 Standard. Marketplace. Do not paste fal Fast Veo over Google Fast.
6. evolink Kling: evolink.ai/kling-3. Marketplace. Ignore their “official Kling $0.089” line unless Kling’s own page has moved to that number. 1080p audio-on is the printed $0.159, not 2 × $0.08.
7. MiniMax points: Video packages. PAYG dollars: pricing-paygo. Use PAYG as list. The 5% package figure is an inference, not a price.
8. Black Forest Labs: docs.bfl.ml/quick_start/pricing and FLUX 3 overview for the 20-second cap.
9. Runway: Dev pricing. Confirm gen4.5 at 12 credits/s and whether a duration cap is now printed on that row; until it is, the 10-second cap here is lotgen’s own test of the route.
10. xAI: grok-imagine-video, grok-imagine-video-1.5 and the video capabilities page (up to 15s). Confirm the 1.5 split and the cap.
11. OpenAI Sora (model on fal/evolink): OpenAI API pricing. Sora 2 $0.10/s at 720p on the day checked.
12. Alibaba: Model Studio pricing and text-to-video API. International wan2.7-t2v is $0.10/s 720P and $0.15/s 1080P. Duration is an integer from 2 to 15 seconds. Do not paste evolink Wan as Alibaba official.
13. Meta ad specs: Ads Guide, Facebook Feed video. TikTok’s own ads-spec page was not reachable when this was checked; read it there before you plan a Reels cell.

Recompute: $/s × billed seconds × attempts × approved variants. If the billed seconds are not a length that provider accepts, stop and check the model id.

Changelog of this page

Related on this site: how to make UGC ads with AI · Kling vs Veo cost · Kling pricing explained · how much AI video generation costs · Veo 3 cost per second · AI video cost calculator · seats and pricing · request an invitation

Request an invitationSeats and pricingA dollar quote on your own keys before a 50-variant board runs.

FAQ