PlayHT pricing

About

PlayHT — later rebranded PlayAI — was a generative voice AI company building text-to-speech, instant voice cloning, and a voice-agent platform. A Y Combinator W23 company founded around 2020–2021 in Palo Alto by Hammad Syed and Mahmoud Felfel, it became known for ultra-realistic voices, 140+ languages, and a developer API (the PlayDialog conversational model and the low-latency Play 3.0 mini) used by teams at Amazon, IBM, and Salesforce. In November 2024 it raised $21M in seed/pre-seed funding led by Kindred Ventures and 500 Global (with Race Capital, Y Combinator, Soma Capital, Pioneer Fund, and TRAC).

In July 2025, Meta acquired Play AI (terms undisclosed). The entire team joined Meta to work on AI Characters, Meta AI, wearables, and audio content, and the standalone product was wound down — the play.ht and play.ai sites no longer resolve. The pricing documented here reflects the structure in place immediately before the acquisition, reconstructed from third-party pricing trackers because the first-party pages are no longer reachable.

For context on the acquirer’s direction, see Meta acquires voice startup Play AI.

Pricing summary : How PlayHT’s pricing model worked

PlayHT priced on characters and words consumed, not minutes of audio — a hybrid of metered subscriptions and a pure-usage developer API. The Free tier gave roughly 1,000 characters/month (some trackers cite up to ~12,500) with all voices and one instant voice clone, but only for non-commercial use with attribution and no API access. Creator ran about $31.20/month billed annually ($374.40/year; historically $39/month month-to-month) and unlocked roughly 3 million characters/year, 10 instant voice clones, and a commercial license. Unlimited removed the character cap at $49/month on a limited-time offer ($99 regular), adding high-fidelity clones and premium voices. Enterprise was custom-quoted and was the tier that unlocked API access, voice cloning at scale, SSO, and team features.

The metering twist: paid plans weren’t truly unlimited until the top consumer tier — Creator charged $4 per additional 10,000 characters once the annual allotment ran out. The developer API was pure per-character usage, roughly $0.15–$0.30 per 1,000 characters at low volume, falling toward custom rates at high volume (high-volume comparisons put effective rates around $0.03–$0.08 per 1,000 characters).

What makes this different: PlayHT metered the unit of text (characters/words) rather than minutes of generated audio — intuitive for creators pasting scripts, but it made cost hard to predict for voice agents where speech length varies. The annual-billing discount (up to ~50%) and a 20% cut for students, educators, and non-profits pushed buyers onto yearly commitments. The whole structure became moot in July 2025 when Meta absorbed the team.

Pricing by product

Sales motions across products: self-serve PLG for Free/Creator/Unlimited (instant signup, annual discount), and sales-led for Enterprise and API/voice-agent deals. Targeted discounts: ~50% off with annual billing and 20% for students, educators, and non-profits (eligibility on application).

Hidden costs : What PlayHT users actually paid

The headline subscription was only the floor. Real PlayHT bills were shaped by character overage and the API, both metered per unit of text. Because the first-party pages are gone, the figures below are reconstructed from third-party trackers and should be treated as illustrative of the pre-acquisition structure.

Other things to budget for: the API was gated to Enterprise/paid tiers, so developers couldn’t build on the free plan; high-fidelity (“ultra”) voice cloning was limited to Unlimited and above; and the annual-billing discount meant the attractive monthly prices required a yearly commitment up front. Voice-agent / PlayDialog conversational usage layered additional per-character (and effectively per-minute) cost on top of TTS.

Want to estimate your own PlayHT bill? Use the PlayHT pricing calculator to model your costs based on usage patterns.

Pricing evolution : PlayHT pricing history and changes

Cadence

Tracked range: 2023–2025. First-party pricing pages (play.ht / play.ai) no longer resolve after the July 2025 Meta acquisition, so history is reconstructed from third-party pricing trackers and acquisition coverage rather than live or Wayback first-party captures.

Notable changes

• 2023 — Launches as a YC W23 generative voice platform: free tier, character-metered Creator/Pro subscriptions, and an ultra-realistic voice API.
• 2024 H1 — Settles on Free / Creator (~$39) + per-character API on the Play 3.0 model family; instant voice cloning from a short sample.
• 2024-11 — Raises $21M (Kindred Ventures, 500 Global, others), ships PlayDialog conversational TTS and a voice-agent platform, and rebrands toward PlayAI.
• 2025 H1 — Consumer pricing lands at Creator $31.20/mo (annual), Unlimited $49 promo ($99 regular), Enterprise quoted; $4 per 10,000-character overage; 20% off for students/non-profits.
• 2025-07 — Meta acquires Play AI; team joins Meta, standalone product wound down, sites stop resolving.

What’s unique : PlayHT’s distinctive pricing mechanics

1. Metered on text, not audio time. PlayHT charged per character/word rather than per minute of generated speech — intuitive for creators pasting scripts, but a poor fit for voice agents where the same prompt can produce wildly different audio lengths. Most rivals (and the voice-agent category PlayAI later chased) price per minute.

2. Annual-billing as the real list price. The attractive Creator number ($31.20) was the annual-billed rate; the month-to-month price was closer to $39, and Unlimited’s $49 was a limited-time promo against a $99 list. The discount structure pulled buyers onto yearly commitments.

3. API behind the paywall. The developer TTS API and high-fidelity cloning sat in Enterprise/paid tiers, so the free plan was a pure consumer funnel — developers had to convert before they could build, the opposite of a developer-first PLG motion.

Strengths & weaknesses

Billing UX : PlayHT billing controls and transparency

• Billing controls — Self-serve signup and plan selection with monthly or annual billing; a refund was available within 24 hours if usage stayed under ~5,000 characters. Plan switching was self-service for consumer tiers.
• Usage visibility — Character consumption was shown against the plan allotment in-product; overage ($4 per 10,000 characters) accrued automatically once the allotment was exhausted, which made spend on heavy projects harder to anticipate than a flat cap.
• Payment options — Card-based self-serve checkout for Free/Creator/Unlimited; Enterprise and API deals were invoiced under a custom contract. After the Meta acquisition, no new purchases are possible — the billing surfaces are gone.

Strategic wins : Why PlayHT’s pricing decisions worked

1. A clean consumer ladder funneled creators

Free → Creator → Unlimited was an easy-to-grasp ladder for content creators, and instant voice cloning on Creator gave a concrete reason to upgrade past the non-commercial free tier. The character allotments mapped to “how many scripts can I narrate,” which creators could reason about. See how AI companies structure pricing.

2. Annual billing anchored revenue ahead of the raise

Pushing the attractive prices onto annual billing (up to ~50% off) pulled cash forward and reduced churn — useful signal heading into the $21M 2024 raise. Related: outcome-based pricing trends.

3. A per-character API opened a second, scalable revenue line

Layering a usage-based developer API and a voice-agent platform on top of consumer subscriptions gave PlayAI a path beyond creator seats into infrastructure spend — the exact capability (natural voices + an easy voice-creation platform) Meta cited when it acquired the team. See choosing the right usage metric.

Areas to improve : Gaps in PlayHT’s pricing approach

1. Character metering didn’t fit voice agents

As PlayAI pivoted toward conversational voice agents, charging per character clashed with how that market prices (per minute of conversation). The unit of value drifted from the unit of billing, making cost hard to forecast for agent builders. See bill shock and cost unpredictability.

2. Promo-versus-list opacity

Listing Unlimited at $49 “limited-time” against a $99 regular price, and Creator at an annual-only $31.20, blurred the true cost. Buyers comparing tools at a glance saw numbers that didn’t hold month-to-month — a transparency gap.

3. Developers locked out of the free tier

Gating the API behind paid/Enterprise tiers meant the free plan couldn’t seed developer adoption, the opposite of the API-first PLG motion that worked for category peers — a missed funnel for the platform’s most scalable revenue line.

Key takeaways

1. Match the billing unit to the value unit. PlayHT metered characters while moving into a per-minute voice-agent market — a drift that makes cost hard to predict and pricing hard to defend.
2. Annual discounts pull cash forward but blur the real price. Creator’s headline $31.20 was annual-only; the month-to-month reality was higher, which can erode trust if not surfaced clearly.
3. Gate the API and you cap developer-led growth. Keeping the API behind paid tiers turned the free plan into a consumer funnel, not a developer funnel.
4. Strong logos and a tight model are acquisition fuel. PlayAI’s natural voices and easy voice-creation platform — plus a $21M raise — made it an attractive acqui-hire for Meta.
5. Standalone voice tooling is consolidating. Meta’s July 2025 acquisition and wind-down of Play AI echoes a broader pattern of independent AI-voice and tooling startups being absorbed by platform giants.

UBP implications

1. Pick a meter your buyer already counts in their head. Characters work for narration but not for conversational agents — the right value metric depends on the workload, and a mismatch undermines predictability. See choosing the right usage metric.
2. Hybrid (subscription + usage API) needs a developer on-ramp. A usage-based API only compounds if developers can start free; gating it behind paid tiers caps the flywheel.
3. Consolidation is a pricing-power signal. When a category’s independents (Play AI → Meta) get acquired and wound down, durable pricing power increasingly sits with platforms, not point tools. See usage-based pricing strategy.

Sources

• Meta acquires voice startup Play AI — TechCrunch (accessed 2026-06-09)
• Meta acquires PlayAI — MENAbytes (accessed 2026-06-09)
• Complete Play.ht Pricing Guide — Voice.ai (accessed 2026-06-09)
• Play.ht pricing and packages for 2025 — Alternatives.co (accessed 2026-06-09)
• Text-to-Speech API pricing compared — Awesome Agents (accessed 2026-06-09)
• PlayAI raises $21M seed — SiliconANGLE (accessed 2026-06-09)
• PlayAI — Y Combinator (accessed 2026-06-09)
• First-party play.ht / play.ai pricing pages — unreachable on 2026-06-09 (post-acquisition wind-down); live capture failed with ERR_NAME_NOT_RESOLVED / ECONNREFUSED.

Bottom line

PlayHT (rebranded PlayAI) was a YC W23 generative voice company that priced text-to-speech and voice cloning on characters/words — a free non-commercial tier, a ~$31–39/month Creator plan (≈3M characters/year), a $49–99 Unlimited plan, and a quoted Enterprise tier that unlocked a per-character developer API and voice-agent platform. Paid plans charged $4 per extra 10,000 characters, and annual billing was the real list price. After a $21M seed in late 2024, Meta acquired Play AI in July 2025; the team joined Meta and the standalone product was wound down, so this entry documents the pricing as it stood just before the acquisition. Browse the pricing blueprint for more fully-researched company profiles.

Want to compare PlayHT against other generative-voice companies? Browse the pricing blueprint.