ElevenLabs' $22 Tier: 121,000 Credits or 220,000 Characters? Both, Says ElevenLabs

Direct answer

Both numbers are live on elevenlabs.io, read the same day (Aug 13, 2026), for the same $22 Creator tier: elevenlabs.io/pricing grants "121,000 credits" a month, while elevenlabs.io/pricing/api prints "220,000 characters of TTS (Multilingual v2 / v3) included ~220 min." The API column is exact arithmetic: every tier's character figure equals the plan price divided by the rate card's own $0.10 per 1,000 characters ($22 → 220,000, $99 → 990,000, $299 → 2,990,000, $990 → 9,900,000). The subscription pool is the number your studio account actually meters. Since a credit corresponds to roughly one character on standard TTS, the same dollar buys 45% fewer subscription-pool characters than the API rate card implies — a credit priced at 1.65-1.82× the list rate, worst at Creator.

Last verified

Aug 13, 2026

How we know

Primary sources plus cited bench data

Sources

2 linked

Evidence used

• elevenlabs.io/pricing read Aug 13, 2026: Creator $22/mo, "121,000 credits" ($11 first month at the time of reading); Pro $99 600,000; Scale $299 1,800,000; Business $990 6,000,000
• elevenlabs.io/pricing/api read Aug 13, 2026 (same session): $22 tier row "220,000 characters of TTS (Multilingual v2 / v3) included ~220 min"; 990,000, 2,990,000 and 9,900,000 on the higher tiers; "$0.10 per 1,000" characters printed as the standalone rate on the same page
• Division check, re-run for this page: 22 ÷ 0.10 × 1,000 = 220,000; 99 → 990,000; 299 → 2,990,000; 990 → 9,900,000 — every API-page figure is exact to the dollar
• Our measured narration pacing, from five published renders (Jul 3, 2026): 826-915 characters per finished minute by voice — used only in the minutes section, and labeled as ours

How we checked this

• Both vendor pages were read in the same session on Aug 13, 2026, so the two figures cannot be explained by a price change between reads.
• The 1.65-1.82× conclusion is derived (division of two vendor-published figures), not measured; the table shows the arithmetic so you can re-run it.
• We do not assert why ElevenLabs presents the same tier in two units — only what each page prints, dated. If either page changes, the change postdates Aug 13, 2026.

Is ElevenLabs Creator 121,000 credits or 220,000 characters?

Depends which ElevenLabs page you read. elevenlabs.io/pricing: Creator $22/month, 121,000 credits. elevenlabs.io/pricing/api, same tier, same price: "220,000 characters of TTS (Multilingual v2 / v3) included ~220 min." Both live, both read Aug 13, 2026. On standard TTS a credit corresponds to roughly one character, so the two figures describe the same resource — 99,000 characters apart.

This is not a stale-cache artifact or a price change caught mid-rollout. We read both surfaces in the same session, and the numbers have been stable across our reads since July. The pricing page speaks the studio subscription's language — a monthly pool of credits, topped up on renewal, rolling over up to two months' worth. The API page speaks rate-card language — characters, priced per thousand.

The row that matters is explicit about the model class: the 220,000 figure is labeled for Multilingual v2 / v3, the same model family our published bench renders use. So this is not a cheap-model-versus-flagship-model footnote; both pages are describing flagship text-to-speech.

Where 220,000 comes from: one division, exact at every tier

Every character figure on the API page equals the tier price divided by the rate printed on the same page — $0.10 per 1,000 characters. $22 → 220,000. $99 → 990,000. $299 → 2,990,000. $990 → 9,900,000. Exact to the dollar, all four tiers. The subscription pages grant 121,000, 600,000, 1,800,000 and 6,000,000 credits at those same prices — 55-61% of the rate-card figures.

Run the division yourself; it never misses. That tells you what the API column is: the plan price restated at the standalone pay-per-use rate, not an independently set allowance. The credit pools on the pricing page are what a subscription actually grants — they are the numbers our own account's usage meter has always matched.

Put the columns side by side and a second pattern appears: the gap is not constant. Creator's pool is 55% of its rate-card figure; Pro, Scale and Business all sit at 60-61%. In per-character terms, a Creator subscription prices a credit at 1.82× the API list rate, and every higher tier at about 1.65×. Creator — the tier most individual buyers land on — is the worst-priced seat in the house on this axis.

Is the ElevenLabs subscription cheaper than the API?

Per character, no — the subscription pool prices a character at 1.65-1.82× the API page's list rate. But the two are not selling the same thing: the subscription buys the studio, the voice library, the commercial license tier and credit rollover; the rate card is a metering price. Decide on what you actually run, not on the headline character count.

If your workload is scripted narration through the studio — the way our bench runs work — the pool is the number to plan on: 121,000 credits is 132-147 finished minutes a month at our measured 826-915 credits per minute, before retakes. The 220,000 figure will never appear in your usage meter, and budgeting on it means running out 45% early.

If you are integrating TTS into a product and metering by volume, the rate card is the honest comparison point, and the division above is your unit price. Either way, the mistake this page exists to prevent is mixing the units: quoting the API page's characters against the subscription's credits produces plans that look 1.8× bigger than they are. Third-party roundups do this constantly, in both directions.

One more discrepancy: ~220 min assumes a pace none of our voices hit

The API page's "~220 min" annotation implies 1,000 characters per finished minute. Our four measured voices — the same 991-character script, identical settings — read 826-915 characters per minute, all slower than the vendor's assumption. At measured pacing, 220,000 characters is 240-266 finished minutes, not 220.

This one is in the buyer's favor, which is exactly why we publish it with the raw audio: a conversion this convenient (1,000 per minute, every tier a round number of minutes) is a rounding convention, not a measurement. If you plan capacity on the vendor's minute estimates you will under-buy slightly; if you plan on its character-versus-credit presentation you will over-estimate a subscription by 45%. Plan on the credit pool and our measured range — both are numbers someone actually metered.

Related ToolProven pages