How AI-creator economics differ from human ones at the core

With a human influencer, pricing is built outside-in: the brand takes the creator’s base rate per post and layers modifiers on top — usage rights, exclusivity, whitelisting. That is per-post pricing with a clear “one airing = one transaction” logic.

Under an AI-creator’s hood is a different model: an amortized setup plus the running cost of generative tooling plus the labour that directs it. The character is built once (bio, visual references, voice, reference set), and each subsequent clip decomposes into generation credits, voice synthesis, and post-production.

The credit side is the only part of that stack with public list prices, and it is smaller than most quotes imply.

Google’s Gemini API price list puts Veo 3.1 video generation at $0.40 per second on Standard (720p and 1080p, audio included) and $0.10 per second on Fast at 720p (ai.google.dev/gemini-api/docs/pricing, checked 17.08.2026). At those list rates, the raw generation for one 30-second clip is $12 on Standard, $3 on Fast — arithmetic on the published per-second rate, before any retries.

ElevenLabs list pricing for voice: Creator $22/month, Pro $99/month, with Starter at $6, Scale at $299 and Business at $990 (elevenlabs.io/pricing, checked 17.08.2026). Multilingual synthesis sits inside those plans; there is no separate “localisation” SKU to budget for.

against your own quote. If raw generation for a clip lands in the single-to-low-double digits of dollars and the quote for that clip is four figures, the difference is not “AI costs” — it is direction, iteration, brand-look conformance, rights, and margin. That is the part worth negotiating, and it is the part a single-number quote hides.

The one disclosed earnings case

The strongest public data point for calibrating expectations is Aitana López, the Spanish virtual model run by The Clueless agency. Per Euronews (page first published 22.11.2023, updated 27.12.2024): “The virtual model can earn up to €10,000 a month, according to her creators, but the average is usually around €3,000”; “She earns just over €1,000 per advert”; “In a year and a half, she has gained more than 343,000 followers on Instagram”; and she had become the face of Big, a sports supplement company. The article attaches no date to the figures themselves, and the source has not re-stated them since the December 2024 update.

Put that one case next to the human grid below and the intuition behind “AI is cheaper” gets a shape: a persona with 343K followers earning ~€1,000 per advert sits far below the band Meltwater publishes for a human at the same follower count — its “Macro (100K–500K)” row, $5,000–$10,000 per Instagram post. But this is one disclosed case against a rate card, not a like-for-like benchmark — different market, different year, creator-side earnings rather than brand-side invoice. It is a reason to ask for a breakdown, not a coefficient to plug into a model.

The seven cost lines: what is on the invoice

Below is each line, what it covers, and what is actually published about its price. Note what the last column does not contain: there is no primary source that prices these modifiers for AI creators specifically, so we do not print numbers there.

Table 1. Integration cost lines and what is published about each

The reason the AI column has no numbers is not modesty — it is that no one has published them. The structural argument for why AI-side modifiers should be lower is real (below), but an argument is not a benchmark, and we are not going to print one as if it were.

1. Production

This is the heart of the quote — everything else is a modifier on it.

For human creators, production pricing is tier-driven, and the two most-cited public grids are in Tables 2–3 below. For an AI creator, production decomposes into two very different halves:

• Credits, which have public list prices — $0.40/sec (Veo 3.1 Standard) or $0.10/sec (Fast, 720p) for video, $22–$99/month for voice. Cheap, predictable, auditable.
• Everything else — character consistency across scenes, brand-look conformance, lore, set design, retakes when the model drifts, and the human hours that direct all of it. Not published by anyone, and the actual driver of a four-figure quote.

Practical takeaway: when a studio quotes production, ask which half the number is. A quote that cannot separate credits from direction hours is a quote that has not been costed.

2. Talent fee

For a human, this is a shoot-day line: presence, rehearsals, re-shoot availability. It is normally folded into the per-post rate in the published grids rather than itemised.

The AI analogue exists but is spread thin: it is the amortization of the character setup (bio, visual references, voice, reference set) across the deliverables that character will produce. Per clip, that is a real cost — but the number depends entirely on how many deliverables the setup is spread over, and no public benchmark quantifies it.

Practical takeaway. Ask the studio for two figures: across how many deliverables the setup is amortized, and what the running cost per period is. If the talent-fee line on your quote reads like a shoot day, that is a tell that the studio is pricing in human-economics logic and has not costed its own pipeline.

3. Usage rights

The most consistently underrated line, and the one where the published number is lower than the market folklore.

Impact.com’s guide states: “Most influencers charge an additional 20 to 50 percent of their base rate for usage rights” (impact.com). If you are being quoted well above that corridor, ask what specifically the extra buys — channels, territories, duration.

Lumanu’s guidance is about a different object and is often misquoted as if it were the same one: “We recommend putting aside 0-100% of the flat rate an influencer gives you so you’re prepared to pay any licensing fee they respond with. The ‘Included’ rates will be closer to 0% and ‘Additional Fees’ will range from 50-100%” (lumanu.com). Name the object precisely: Lumanu’s 0–100% reserve and 50–100% “Additional Fees” sit under the heading “What you should expect to pay for whitelisting” — they describe whitelisting licensing, not usage rights and not reposting. That is a budgeting reserve for an additional licence, not the typical modifier — the two get glued together into an inflated “50–100% typical” figure across the industry, and they should not be.

On perpetual rights, the honest position is qualitative. Impact.com’s own wording: “If a brand requests perpetual rights, charge a premium—you’re giving away permanent control of your work”, and “You should charge a significantly higher rate”. No public source we checked puts a multiplier on it — so treat “perpetual doubles the price” as a negotiating claim, not a benchmark, from whichever side of the table it comes.

For AI creators the modifier should be structurally lower, because the studio can generate new variants at near-zero marginal cost, so the “one image, one brand” scarcity that justifies the human premium is softer. That is a reason to push back on a human-sized rights modifier — not a published discount you can cite.

4. Exclusivity window

A ban on competitor work for an agreed term.

Impact.com: “Most creators charge between 20 and 100 percent of their base rate, depending on the length of exclusivity and the brand’s size” (impact.com, checked 17.08.2026). The spread inside that corridor is set by scope and duration, which is exactly why scope has to be written down: a category exclusive in the shape “no beauty-category competitors for 90 days” is a different product from a market-wide one, and should not carry the same percentage.

For AI creators the mechanic is unchanged — a persona locked to your category cannot work for a competitor either. What differs is the studio’s alternative: with a portfolio of personas, exclusivity on one character costs the studio less than exclusivity on a person costs that person. Again: a negotiating argument, not a published number.

5. Whitelisting

The brand runs paid ads under the creator’s native handle. Two pricing models are published, and they are not equivalent:

• Flat share of the deal: “A reasonable rate is 25% of the base deal size per month for every 30 days the brand boosts your native post” (topgrowthmarketing.com).
• Share of ad spend: “Creators typically charge five to 20 percent of the brand’s total paid ad spend used to promote the content” (impact.com). Lumanu describes the same model with a worked example — “the influencer might charge 4% of ad spend”.

The choice between them matters more than the percentage: a flat share is predictable, a share of spend scales with your media budget and can quietly become the largest line on the invoice if the campaign works.

Whitelisting is also not universal: per Lumanu, “51% of influencers said they charge a fee to allow partners to whitelist, boost, or amplify content above and beyond their usual fees for content creation” (lumanu.com) — so roughly half do not charge separately at all, and it is worth asking before assuming the line exists.

The rate logic is the same for AI and human creators, because the mechanic is the same: paid amplification through a handle.

6. Repost / cross-platform rights

If the brand wants to re-publish the clip on its own channels (brand IG account, YouTube, corporate TikTok), that is a licensing question, not a new production. In the published guidance it usually lives inside the usage-rights licence rather than as its own line, and we found no primary source that prices reposting separately — the Lumanu ranges quoted above belong to whitelisting, not to this line, and should not be carried over to it.

AI economics are simpler here because a repost takes nothing away from the persona: it does not lose the ability to work with the same image elsewhere. Whether a given studio prices it that way is a question for the quote.

7. Agency margin

When working through an influencer agency or an AI-studio intermediary, there is a margin. We could not find a single primary source that publishes a range for it, and we are not going to invent one — the honest instruction is procedural rather than numeric:

• ask for the margin as an explicit line rather than a percentage embedded in the others;
• ask what it buys — casting, negotiation, production supervision, reporting, media handling, liability;
• compare the answer to what your own team would otherwise do.

An intermediary that cannot name its margin is telling you where it lives: inside the other six lines.

What the published rate cards say — and why they do not line up

Two vendor guides dominate the citations in this category. Both are live and both are internally consistent. What they do not share is a vocabulary: the same tier name covers different audiences in the two guides, which is the single most useful fact in this section. The tables below are therefore aligned by follower band, not by tier label, with each guide’s own label kept in the cell.

Table 2. Instagram per-post benchmarks by follower band (human creators)

Sources: meltwater.com, shopify.com — both checked live 17.08.2026. Meltwater and Shopify define the tiers differently: Shopify’s “macro” is 501K–1M, Meltwater’s is 100K–500K — so a quote benchmarked against “macro” means two different things depending on which guide the seller read. Meltwater’s top band stops at 500K+, which is why its mega row spans the last two rows here.

Table 3. YouTube and TikTok per-post benchmarks by follower band (human creators)

Sources: meltwater.com, shopify.com, checked live 17.08.2026.

Look at the tier labels before you look at the numbers. The disagreement between these two grids is mostly definitional rather than numeric. Aligned on the same follower band, they are closer than the labels suggest: on TikTok at 100K–500K, Meltwater publishes “$1,250-$2,500” and Shopify “$1,200 to $5,000 per post” — the same floor, a 2x difference at the ceiling. On Instagram at that band it is “$5,000-$10,000” against “$1,600 to $10,000” — the same ceiling, a wider floor. Compare by label instead of by band and you get an apparent 4–6x gulf that is really just two different audiences wearing the same word.

Neither the label mismatch nor the remaining spread is a mistake in either source: the guides aggregate different deal populations and different geographies, and define what one “post” includes differently. The practical consequence is the point of this whole article — there is no “market rate” to anchor on. The published corridors are wide on their own terms (Meltwater’s 10K–100K Instagram row alone runs from $100 to $5,000), and the tier vocabulary on top of them is not standardised. Inside that corridor your price is set by which components you buy and how well you specify them. A quote is not validated by matching a benchmark; it is validated by adding up — and any quote justified as “the macro rate” should be made to name its follower band first.

Why there is no “AI discount” number — and what to use instead

The tempting sentence to write here would be “at comparable reach, an AI creator lands X% below these rates.” We checked the sources that circulate for that claim, and none of them survive: the numbers either are not on the pages that are cited for them, or come from domains that do not resolve at all. So the sentence does not get written.

What can be said, in decreasing order of confidence:

1. Generative tooling has published list prices, and they are low relative to the quotes built on them ($0.40/sec Veo 3.1 Standard; $22–$99/month for voice — see above).
2. The one disclosed AI-persona earnings case (Aitana López, ~€1,000 per advert at 343K+ followers, as reported by Euronews and not re-stated by the source since its December 2024 update) sits below Meltwater’s published 100K–500K human rate card for the same follower count.
3. The structural reasons for lower modifiers on the AI side — near-zero marginal cost of variants, portfolio-level rather than person-level scarcity — are real, and they are legitimate negotiating arguments.
4. Not established. Any specific multiplier or percentage discount per deliverable, in either direction.

The practical replacement for the missing multiplier is arithmetic on your own campaign:

That formula makes the actual decision visible. If your campaign is one or two deliverables, the setup does not amortize and the AI route will not produce dramatic savings. If your campaign is a series of ten or more, especially localized across markets, the amortization does the work — and the localisation line is genuinely small, because multilingual synthesis is inside the same voice plans quoted above rather than a separate purchase.

The final integration price formula

Integration price = Base content fee × (1 + usage-rights modifier + whitelisting modifier + exclusivity modifier + production-complexity modifier) + agency margin.

That formula holds for both AI and human creators. The difference is in the absolute values of the coefficients, and in the fact that for AI the base content fee decomposes not into “an hour of creator time” but into credits plus direction hours.

Our practice. We ask for every quote in this shape — seven lines, each with a scope and a term — before discussing the total. The exercise is worth doing even when you accept the total unchanged: it tells you which lines are negotiable next time.

Three worked examples

These are illustrative constructions, not deals we have run. The base fees are chosen to sit inside the published corridors above; the modifiers are taken from the published corridors and labelled. Nothing here is a forecast of revenue — the only honest projection of return is one built on your own conversion data.

Example 1. D2C brand, micro-tier AI creator on Instagram, performance campaign

• Base production (1 Reel + 3 Stories): $1,200 — inside the micro corridor for Instagram ($100–$5,000 Meltwater / $250–$5,000 Shopify).
• Usage rights, brand repost + paid usage, 60 days: +50% = $600 — top of Impact.com’s published 20–50% corridor.
• Exclusivity, 30 days, direct competitors only: +20% = $240 — bottom of Impact.com’s published 20–100% corridor, appropriate for a narrow scope and a short term.
• Whitelisting, 30-day boost: +25% of base = $300 — Top Growth Marketing’s published rate.
• Total: $2,340.

What to check before signing: whether the whitelisting line is flat (as modelled here) or a share of ad spend. At Impact.com’s published 5–20% of spend, a $25,000 media budget behind this content would produce a $1,250–$5,000 line instead of $300 — larger than the production fee.

Example 2. Mid-market brand, mid-tier AI creator, YouTube integration

• Base production (dedicated integration video): $6,000 — inside Meltwater’s YouTube micro/mid corridor ($200–$10,000).
• Usage rights, broad and long, for cuts in paid: +50% = $3,000 — top of the published corridor.
• Exclusivity, 90 days, category: +50% = $3,000 — mid-corridor, priced for the longer term.
• Total: $12,000.

Now the honest CPM check, because this is where deals of this shape usually fail. At 60,000 views, $12,000 is a $200 CPM — above the $50–$150 corridor that creator integration bundles command (see the CPM section below). To land at $100 CPM you need 120,000 views. So either the view floor goes into the contract, or the deal has to be justified by measured lift rather than by CPM. “The creator usually gets around 60K” is not a number you can sign against.

Example 3. Enterprise brand, mixed campaign, hero AI creator + 10 nano placements

• Hero placements: no public rate card exists for hero-tier AI personas, so the anchor is the human mega band — Shopify’s “Mega: $10,000 to more than $50,000 per post”. Modelled at $10,000 × 3 deliverables = $30,000.
• 10 nano AI placements at $150 each (inside Shopify’s Instagram nano band of $25–$150) = $1,500.
• Usage rights aggregated on the hero deliverables: +50% = $15,000.
• Production uplift (custom wardrobe, lore tie-in, extra scene consistency work): $5,000.
• Total: $51,500.

Cross-check that against the market: per Inbeat’s compilation, “in 2023, brands invested an average of $18,000 to $30,000 per virtual influencer campaign” (inbeat.co) — so a hero-led structure like this one sits well above that reported 2023 average, which is exactly what you would expect and exactly what you should be able to justify internally before committing.

Why a creator integration costs more per thousand than paid inventory

The natural pushback from the brand: why pay a creator a $50–100 CPM when the same reach is available on the paid auction at a fraction of that? Short answer — it is not the same purchase. Long answer, in published numbers, all from the same benchmark page (influencers-time.com, 2025):

• YouTube TrueView (skippable pre-roll): $10–$30 CPM on the auction, $25–$45 on reserve.
• Meta Feed and Reels: $8–$22 CPM.
• Creator integration bundles on YouTube: CPMs between $50 and $150 — that is the CPM level of the bundle, not a premium added on top of a paid-social CPM.

Verbatim from that source: “A $100 CPM YouTube creator integration that includes custom production, organic reach, and a creator’s audience trust is not the same purchase as a $15 CPM YouTube pre-roll.”

In a creator integration the brand is not buying an impression — it is buying custom production, organic reach on top of whatever is paid, and the creator’s standing with their audience. A multiple over raw inventory is normal. What is not normal is a multiple you cannot decompose, which brings us back to the seven lines.

Negotiation checklist: what to ask the agency or the platform

• [ ] Ask for the quote as seven lines (production / talent / usage rights / exclusivity / whitelisting / repost / agency margin), not a single number.
• [ ] On each modifier, clarify: a percentage of what, for what term, for what scope (category? market? platform?).
• [ ] For an AI creator, ask separately: across how many deliverables the setup is amortized, and what running cost is included in the quote.
• [ ] Ask which usage rights are included by default, and what dropping perpetual is worth — the seller is entitled to a premium for permanent rights, but no published benchmark sets it, so it is negotiable on both sides.
• [ ] For whitelisting, establish which model applies: a flat share of the deal (25% per 30 days is the published benchmark) or a share of ad spend (5–20% is published) — and model the second one against your actual media budget.
• [ ] Lock down which KPIs are measured and who pays for measurement (this line often hides inside production).
• [ ] Put a view or impression floor in the contract for any deliverable whose economics you justified with a CPM.
• [ ] Ask the intermediary to name its margin and say what it buys.
• [ ] Ask for a comparable campaign from the last 12 months — if the studio cannot produce one, that is a signal.

What the campaign-level numbers actually show

A few figures worth having ready when you defend the budget internally — with their limits stated, because the limits are the useful part.

, with a caveat. Per Inbeat’s compilation of virtual-influencer statistics, “in 2023, brands invested an average of $18,000 to $30,000 per virtual influencer campaign” (inbeat.co). This is a 2023 aggregate from a secondary compilation, not a 2026 figure and not a forecast, and the underlying dataset is not named on the page.

The same page reports an “average ROI of 13.7%, slightly exceeding the 12.3% ROI of campaigns featuring human influencers”. A 1.4-point difference at an unstated sample size is close to noise, and the source does not name the study behind it — so “AI beats human on ROI” is too strong a claim to carry into a budget meeting. The defensible formulation: at comparable budgets, the reported figures show the virtual channel roughly matching human performance rather than beating it.

A virtual persona incurs no travel, accommodation, or location costs, and no scheduling constraint tied to one person’s calendar. That is a real structural saving on the production line — we have not found a primary source that quantifies it as a percentage, and the percentages circulating in the category trace back to pages that do not contain them.

Frequently asked questions

How much does an AI-creator integration cost on average in 2026?

There is no published median per deliverable. The one campaign-level figure we could verify is Inbeat’s compilation: “in 2023, brands invested an average of $18,000 to $30,000 per virtual influencer campaign” (inbeat.co) — a 2023 aggregate from a secondary source, not a 2026 figure and not a forecast. For a single deliverable, price it from the components rather than from a benchmark.

What components go into an AI-creator integration price?

Up to seven lines: production, talent fee, usage rights, exclusivity window, whitelisting, repost / cross-platform rights, agency margin. The published modifier benchmarks are human-creator ones: usage rights “an additional 20 to 50 percent of their base rate”, exclusivity “between 20 and 100 percent of their base rate, depending on the length of exclusivity and the brand’s size” (impact.com), whitelisting “25% of the base deal size per month for every 30 days the brand boosts your native post” (topgrowthmarketing.com). No equivalent published corridors exist for AI creators — request the quote broken into these lines rather than as a single number.

Is an AI creator cheaper than a human at comparable reach?

No published benchmark quantifies a gap per deliverable, and the figures circulating for it did not survive our source check. What is verifiable: generative tooling has low list prices ($0.40/sec for Veo 3.1 Standard video, $22–$99/month for ElevenLabs voice), and the one disclosed AI-persona case — Aitana López at just over €1,000 per advert with 343K+ followers, per Euronews (page first published 22.11.2023, updated 27.12.2024) — sits below Meltwater’s published 100K–500K human rate card ($5,000–$10,000 per Instagram post, meltwater.com). That is one case against a rate card, not a benchmarked discount.

How much do usage rights and whitelisting cost?

The published corridors are for human creators: usage rights add “an additional 20 to 50 percent of their base rate” and whitelisting runs either “25% of the base deal size per month for every 30 days the brand boosts your native post” or “five to 20 percent of the brand’s total paid ad spend” (impact.com, topgrowthmarketing.com). Perpetual rights carry a premium — Impact.com advises creators to “charge a premium” — but no public source we checked puts a multiplier on it, so treat any specific multiple as a negotiating claim from either side.

Why do the published rate cards disagree with each other?

Because they aggregate different deal populations, define a “post” differently — and, above all, because they do not define the tiers the same way. Shopify’s “macro” is 501,000–1M followers, Meltwater’s “Macro” is 100K–500K, so the same word covers two different audiences. Compared on the same follower band — 100K–500K on TikTok — Meltwater publishes “$1,250-$2,500” and Shopify’s mid-tier row “$1,200 to $5,000 per post”: the same floor, a 2x difference at the ceiling. Use the grids as a corridor check, not as a price, and make any quote justified as “the macro rate” name its follower band first.

How much do AI creators themselves earn?

The most verifiable public figure is Aitana López, The Clueless’s virtual model: “The virtual model can earn up to €10,000 a month, according to her creators, but the average is usually around €3,000”, “She earns just over €1,000 per advert”, on “more than 343,000 followers on Instagram” (Euronews, page first published 22.11.2023, updated 27.12.2024). The article attaches no date to the figures themselves, and the source has not re-stated them since that December 2024 update.

What if the agency sends a quote as a single number?

Ask for the breakdown into the seven lines. A refusal is a signal: either the price carries a margin that would not survive being named, or the studio has not costed its own pipeline and is padding the number as insurance. Either way, the brand has room to negotiate.

When does AI-creator economics not pay off?

When the campaign is one or two deliverables. The character setup is a fixed cost, and across two clips it barely amortizes — you are effectively paying to build a persona you will not reuse. The economics work at full strength on series of ten or more deliverables, especially with multi-market localisation, where multilingual voice sits inside the same subscription rather than adding a separate line.

→ Browse the AI-creator catalogue