Poe pricing

About

Poe is Quora’s multi-model AI chat aggregator: one app (and one subscription) that puts ChatGPT, Claude, Gemini, Grok, plus image, video, and audio models — and tens of thousands of user-built bots — behind a single login. Instead of juggling separate subscriptions to OpenAI, Anthropic, Google, and xAI, you pay Poe once and route each question to whichever model fits.

Poe is owned and operated by Quora, the Q&A company, which raised $75M from Andreessen Horowitz (announced 2023) largely to scale Poe. Poe’s reported revenue contribution roughly doubled from ~$30M in 2024 to ~$65M in 2026, making it Quora’s clearest AI-era growth engine. Poe launched the subscription in early 2023 and added creator monetization later that year.

The pricing question Poe answers is a real one: if you use several frontier models, paying for each provider separately is expensive and fragmented. Poe’s bet is that a single metered subscription — where a flat fee buys a pool of compute points you spend across every model — is simpler and, for mixed usage, cheaper.

For the most current information, visit Poe.

Pricing summary : How Poe’s pricing model works

Poe is subscription plus a shared compute-points meter — a hybrid model. You pick a monthly tier; that tier grants a compute-points allowance; every message, image, video, or audio generation spends points from that allowance at a rate that depends on the model you chose.

The ladder (USD, verified 2026-06-16):

• Free — ~3,000 compute points per day (about 150 short messages), lighter models only.
• $4.99/mo — 10,000 compute points per day (a daily allowance, not monthly).
• $19.99/mo — 1,000,000 compute points per month (the headline plan); $199.99/year (~$16.67/mo).
• $49.99/mo — ~1.65 million compute points per month.
• $99.99/mo — ~3.3 million compute points per month.
• $249.99/mo — ~8.25 million compute points per month (marketed “up to ~12.5M” in some regions/promos).

All paid plans are roughly 17% cheaper billed annually. The genius — and the catch — is that points map to model cost: a lightweight text model costs ~10–20 points per message, while frontier models cost hundreds to thousands, and image/video generations cost the most. So $19.99 stretches to tens of thousands of cheap messages or only a few hundred frontier-video generations.

What makes this different: Most AI apps sell either a flat “unlimited-ish” subscription (ChatGPT Plus) or raw per-token API billing. Poe sits in between: one flat fee, but metered access across many vendors’ models on a single currency. You never see a token bill, yet you still feel usage — when points run low you upgrade or buy more.

Pricing by product

Beyond subscriptions, Poe runs a creator marketplace: bot creators set a per-message price (in points), users pay from their own points budget, and creators are paid in US dollars. Creators also earn up to $20 per new subscriber their bot brings in. An enterprise/business arrangement exists for organizations (custom, sales-quoted — not a public rate card).

Sales motions across products: self-serve / PLG for the entire consumer ladder — every tier is purchasable in-app with a card, no sales call. Enterprise is the only sales-led path.

Hidden costs : What Poe users actually pay

The headline $19.99/mo looks flat, but the real spend is governed by which models you use. Points are the hidden meter.

The trap is that a single high-end video generation can cost more points than thousands of cheap text messages. Users who lean on image/video models or frontier reasoning models exhaust a million points faster than they expect and either upgrade or buy add-on points — turning a “flat” $19.99 plan into variable spend.

Want to estimate your own Poe bill? Use the Poe pricing calculator to model points spend by your model mix.

Pricing evolution : Poe pricing history and changes

Cadence

Tracked range: 2023–present. Points allowances and tier counts have shifted as Poe added higher-throughput tiers and as frontier/video models raised the cost of a “message.”

Notable changes

• 2023-02 — Poe Subscription launches at ~$19.99/mo, establishing the subscription-plus-points model.
• 2023-11 — Creator monetization v1: subscription bounty (up to $20 per new subscriber).
• 2024-04-08 — Creator price-per-message launches (US first); creators paid in USD, users charged in points.
• 2024–2026 — Ladder broadened to six tiers ($4.99, $19.99, $49.99, $99.99, $249.99) plus add-on points, scaling allowances for power users.

What’s unique : Poe’s distinctive pricing mechanics

1. One currency across every vendor’s models. Compute points abstract away the wildly different per-token costs of GPT, Claude, Gemini, Grok, and image/video models into a single budget. You pay Poe once and Poe settles with the underlying providers — a true aggregator pricing play.

2. Subscription that’s secretly metered. The flat tier feels like ChatGPT Plus, but it’s a points budget, not unlimited access. This lets Poe offer frontier and video models (which it could never give away unlimited) on a consumer-friendly flat price, while protecting margin via the meter.

3. A two-sided marketplace baked into the meter. Creator bots charge per message in the same points users already hold, and Poe pays creators in dollars. The points currency is simultaneously the consumer’s spend budget and the creator’s revenue rail — a rare consumer-AI monetization loop.

Strengths & weaknesses

Billing UX : Poe billing controls and transparency

• Billing controls — Self-serve in-app subscription; monthly or annual (annual ~17% off). Upgrade/downgrade between the six tiers; buy add-on points when an allowance runs dry.
• Usage visibility — Points balance is shown in-app, and every bot’s “Info” button displays that model’s exact per-message point cost before you send — unusually transparent metering for a consumer app.
• Creator dashboard — Creators get an analytics dashboard tracking earnings across paywalls, subscriptions, and per-message revenue; payouts are monthly, 30–45 days after period close.
• Payment options — Card-based consumer checkout (and app-store billing on mobile); enterprise arrangements are sales-quoted.

Strategic wins : Why Poe’s pricing decisions worked

1. The aggregator wrapper turns model chaos into one bill

By pricing in a neutral compute-points currency, Poe absorbs the complexity of dozens of providers’ per-token rates and sells one simple subscription. That’s the same first-click simplicity that powers the shift away from per-seat AI licenses — buyers want one predictable entry point, not five vendor invoices.

2. Metered subscription protects margin while feeling flat

A flat fee that secretly buys a usage budget lets Poe offer expensive frontier and video models on a consumer price without unlimited-use blowups. It’s a textbook case of choosing the right usage metric: points track real cost-to-serve while staying legible to non-technical users.

3. Points double as a creator revenue rail

Because creator bots charge in the same points users already hold, Poe built a two-sided marketplace with zero new currency friction — and reports a “tens of millions” annual creator-payout run rate. See outcome-based and marketplace pricing trends for why this matters.

Areas to improve : Gaps in Poe’s pricing approach

1. Points are opaque until you’ve spent them

Even with the “Info” button, mapping “1 million points” to “how many Claude Opus messages or how many videos” is hard for new users. The result is classic bill-shock and cost unpredictability — users hit a wall mid-month and feel surprised.

2. Heavy single-model users can beat the price elsewhere

If you only ever use one model heavily, going direct to that provider (or its native subscription) is often cheaper than Poe’s metered allowance. Poe wins on breadth, not single-model depth — and doesn’t always signal that clearly.

3. Shifting per-message costs erode trust

Because point costs track underlying model prices, the “value” of a tier silently changes when providers reprice. Clearer change communication and per-tier “what you can do” estimators would reduce the feeling that the meter moves under you.

Key takeaways

1. Poe is subscription + a shared compute-points meter — a flat fee buys a usage budget, not unlimited access.
2. The ladder runs Free → $4.99 → $19.99 → $49.99 → $99.99 → $249.99/mo, with ~17% off annual; $19.99/1M-points is the headline plan.
3. Model mix is the real cost driver — frontier and image/video models burn points orders of magnitude faster than lightweight text.
4. Points are a two-sided currency — the same units users spend are how creators earn (paid out in USD, plus up to $20/subscriber).
5. The aggregator value is breadth — one bill for many vendors’ models; less compelling if you only use one model heavily.

UBP implications

1. A neutral usage currency lets you aggregate disparate vendor costs into one clean subscription — points abstract per-token chaos the way credits do in many UBP designs.
2. “Flat fee buys a budget” is a powerful hybrid for AI products: legible like a subscription, margin-safe like usage billing. The cost is predictability, which you must mitigate with visible metering — see our introduction to usage-based pricing for the tradeoffs.
3. The same meter can power both buyer spend and creator earnings, turning a consumption currency into a marketplace rail — a pattern worth studying for any platform building a creator/partner economy on top of usage.

Sources

• Poe pricing page (accessed 2026-06-16)
• Poe subscription plans (accessed 2026-06-16)
• Poe Purchases FAQ — Help Center (accessed 2026-06-16)
• Poe Creator Monetization FAQ — Help Center (accessed 2026-06-16)
• New on Poe: Creator monetization via price per message (accessed 2026-06-16)

Bottom line

Poe sells one of the cleanest aggregator pricing models in consumer AI: a flat subscription ($4.99 to $249.99/mo, headline $19.99/mo) that buys a pool of compute points you spend across ChatGPT, Claude, Gemini, Grok, and image/video models — plus a creator marketplace that pays bot builders in real dollars from that same points pool. The strength is breadth and simplicity; the weakness is that “flat” pricing is really metered usage, and frontier/video models drain the meter fast. Best for people who genuinely use many models; less compelling for single-model power users.

Want to compare Poe against other AI platform companies? Browse the pricing blueprint.