Poe Pricing (2026): $4.99 Wins Until You Need Bursts

Poe starts at $4.99 a month, but its cheapest paid plan is the best-value plan only when your usage stays under 10,000 points every day. The $19.99 tier buys burst capacity, not cheaper points: at current allowances, every pooled paid tier costs about $30.30 per million points before the annual discount.

Poe pricing at a glance

Poe costs from $4.99 to $249.99 a month, with five paid allowances and annual billing at about 17% less. Prices and allowances were verified against Poe's live pricing page on 17 August 2026.

Poe does not need invented tier names to explain the offer. The allowance is the product: one plan resets daily, while the other four give you a monthly pool.

*Unit costs assume every included point is used. The monthly daily-plan calculation uses a 30-day month; the annual calculation uses 365 days.

All five subscriptions buy the same broad class of benefits: substantially more messages than non-paying access, broader model availability, and each bot's maximum input size and chat history. The price difference is primarily capacity, not a progression from a weak product to a fully featured one.

That distinction matters. A founder who sends a steady stream of short prompts has a different cost profile from an operator who runs one document-heavy research batch every Friday. Poe charges both in points, but only the second person needs a pool that survives a busy day.

The $4.99 versus $19.99 decision

The decision flips on your busiest day, not your average month. Poe's $4.99 plan can deliver as many as 300,000 points in a 30-day month, but only in 10,000-point daily slices. Unused capacity expires each day, so yesterday's quiet schedule cannot fund tomorrow's large job.

The $19.99 plan gives 660,000 points in one monthly pool. That is 2.2 times the 30-day capacity at about four times the price. On raw quantity alone, it looks worse. What it sells is the freedom to spend well beyond 10,000 points on a busy Tuesday and almost nothing on Wednesday.

That extra $15 is the burst premium. It pays for timing flexibility.

A solo technical builder who uses 6,000 points most weekdays should stay on $4.99. A senior operator who uploads a long report, asks several frontier models to analyze it, and then generates creative assets in one session can cross 10,000 points before lunch. The same monthly total can fit both people, but only the second workload needs pooled billing.

The clean decision rule is:

• Stay free while the daily grant completes the work and a reset is merely an inconvenience.
• Pay $4.99 when you need more access but can keep every active day at or below 10,000 points.
• Pay $19.99 when any normal workday can exceed 10,000 points and total monthly use stays at or below 660,000.
• Compare add-ons with the $49.99 tier when monthly use exceeds 660,000, especially if the excess is temporary.

There is no economic reason to buy $19.99 because it looks like the standard choice. Buy it only when the daily wall interrupts paid work. If a monthly pool ends with 300,000 unused points, you paid a burst premium for a burst that never happened.

Cost per outcome and the break-even math

Poe's useful unit is cost per million points consumed, not cost per subscription. That normalization exposes two things the sticker prices hide: the daily plan has the lowest unit cost at full utilization, and every monthly pool is priced almost perfectly linearly.

The daily plan is cheap only when you use it steadily

At 10,000 points a day for 30 days, the $4.99 plan delivers 300,000 points. The maximum-use cost is $16.63 per million points. On annual billing, using all 10,000 points across 365 days delivers 3.65 million points for $49.99, or $13.70 per million.

Utilization changes that result quickly. If a solo operator consumes 6,000 points on 20 workdays, the month totals 120,000 points. The cash bill is still only $4.99, but the effective rate becomes $41.58 per million points. That does not make the plan bad. It shows why unit economics and total spend answer different questions.

Use total spend to decide whether the subscription is affordable. Use cost per consumed point to decide whether you are buying too much expiring capacity.

The pooled tiers do not reward volume

The 660,000-point pool costs $30.29 per million points when fully used. The 1.65 million, 3.3 million, and 8.25 million pools each round to $30.30 per million. Poe is not offering a bulk discount as you climb. It is multiplying price and points together.

The annual pattern is equally flat. Every pooled tier costs about $25.25 per million included points when fully used. A mid-market CTO should therefore choose the smallest pool that clears normal demand. Buying 8.25 million points because the team may grow does not improve the rate; it increases the amount that can expire.

Even full-pool calculations can flatter the value. A $19.99 subscriber who uses 500,000 of 660,000 points pays an effective $39.98 per million consumed points. The unused 160,000 points have no residual value at the reset.

Annual billing wins in month 11

Poe displays a 17% annual discount, but the practical break-even is more useful. The $4.99 plan costs $49.90 after ten monthly payments, just below the $49.99 annual charge. The eleventh payment pushes monthly billing above annual. The same pattern holds across all five tiers.

Annual billing therefore makes sense only when you expect to use the same allowance for at least 11 of the next 12 months. The entry plan saves $9.89 over a full year. The $19.99 plan saves $39.89. Those savings are genuine, but small compared with paying for a plan that no longer fits.

Poe fees and final charges are generally non-refundable, subject to its formal policy, applicable law, and the billing platform's rules. That makes the annual option a forecast, not a default. One reversible monthly cycle is worth more than a 17% discount applied to the wrong allowance.

Which Poe tier fits which workload

The right Poe tier is the smallest allowance that survives your normal peak without creating a large expiring balance. Each tier below has a clear job and a clear reason to skip it.

10,000 points a day at $4.99 monthly

The 10K daily tier is the best cash-efficient choice for a disciplined individual. It fits a solo builder, founder, or operator who uses Poe regularly but can spread work across the week.

Its strength is the low entry cost and the best maximum-use point rate. Its wall is timing. A quiet weekend contributes nothing to Monday's balance, and a single expensive model interaction can consume a meaningful share of the day.

Choose it for steady chat, drafting, comparison, and occasional model switching. Skip it when a normal work session needs more than 10,000 points or when deadlines make waiting for the next reset unacceptable.

Annual billing is $49.99, displayed as $4.17 a month. Do not take the annual option until the daily ceiling has survived at least one representative billing cycle.

660,000 points a month at $19.99 monthly

The 660K monthly tier is the first sensible choice for uneven professional use. It fits a funded founder who alternates between light chat and dense diligence, or a senior operator whose large files and model comparisons arrive in batches.

The benefit is one pool that can absorb a heavy day. The cost is a unit rate almost twice the fully used daily plan. It also has no rollover, so buying flexibility without using it destroys the value.

Choose it when the 10K daily limit blocks normal work but the monthly total remains below 660K. Skip it when the daily plan already clears every day, or when 660K is predictably too small and add-on purchases become routine.

Annual billing is $199.99, displayed as $16.67 a month. It is the most direct sticker comparison with ChatGPT Plus, Claude Pro, and Perplexity Pro.

1.65 million points a month at $49.99 monthly

The 1.65M monthly tier is for recurring high-volume work, not a one-time project. It fits a small operating group that repeatedly consumes more than 660K points and wants a predictable recurring allowance.

Its wall is the add-on comparison. The $19.99 plan plus 1 million add-on points at Poe's advertised $30 rate costs the same $49.99 and yields 1.66 million available points, 10,000 more than this tier. Add-on points also remain usable for one year rather than expiring at the monthly reset.

Choose 1.65M when the capacity is needed every month and automatic replenishment is worth more than flexibility. For a one-off campaign, diligence sprint, or migration, keep the smaller plan and buy the longer-lived add-on.

Annual billing is $499.99, displayed as $41.67 a month. The unit rate remains about $25.25 per million included points when fully used.

3.3 million points a month at $99.99 monthly

The 3.3M tier is a power-user budget. It fits a technical lead or operator running frequent frontier-model, media, or API work where 1.65M has become a recurring constraint.

The benefit is capacity without changing the unit rate. The downside is the same: unused monthly points disappear. There is no volume discount to protect a speculative purchase.

Choose it only after Points History shows repeated months above 1.65M. Skip it when usage is project-based, because add-ons preserve optionality for a year.

Annual billing is $999.99, displayed as $83.33 a month.

8.25 million points a month at $249.99 monthly

The 8.25M tier is for sustained throughput, not status. It fits a heavy individual workload or an eligible shared subscription whose recurring activity is already measured in several million points.

At $249.99, the plan still costs about $30.30 per million included points on monthly billing. Nothing becomes cheaper. The buyer is paying for a larger recurring ceiling.

Choose it only when smaller pools and add-ons create predictable operational friction every month. Skip it for a team that has not set ownership rules, because shared-subscription members can each consume up to 100% of the common balance.

Annual billing is $2,499.99, displayed as $208.33 a month. The annual commitment is large enough that two inactive months bring the purchase back to its break-even boundary.

Four buyer workloads and the plan each one should choose

The tier labels become useful only after the workload has a calendar. These four modeled scenarios show why total monthly points alone can recommend the wrong plan.

A solo technical builder with steady weekday use

Suppose a technical founder consumes 6,000 points on each of 20 workdays. That is 120,000 points in a month. The $4.99 daily plan covers the pattern with 4,000 points of headroom on every active day, and its effective cost is $41.58 per million consumed points.

That effective rate is worse than the plan's fully used rate because weekend capacity disappears. It still produces the lowest cash bill. Moving to the $19.99 pooled plan would buy flexibility that this workload does not need and raise monthly spend by $15.

The choice changes if one scheduled job regularly crosses 10,000 points. A daily ceiling is an operating constraint, not just an allowance. The founder should remain on $4.99 while work fits under it, then move only when a normal workday begins to fail.

A funded founder with irregular diligence work

Now model 400,000 points in a month, concentrated around a few diligence or planning days. The daily tier can theoretically deliver 300,000 points across a 30-day month, so it is too small even before burst timing matters. The 660K pool at $19.99 is the right operational answer.

At 400K of actual use, the effective price is $49.98 per million consumed points. That looks less efficient than Poe's headline unit economics, but the unused capacity is the price of removing the daily bottleneck. The founder is buying control over timing.

The mistake would be jumping to 1.65M because one month feels busy. Keep the 660K tier until Points History shows a repeatable shortfall. A workload with irregular peaks should be governed by the smallest pool that clears normal demand, plus an add-on for an exceptional project.

A senior operator with one 1.1M-point project month

For a one-time campaign, migration, or research sprint requiring 1.1M points, pair the $19.99 monthly pool with 1M add-on points at Poe's approximate $30 rate. The total cash cost is $49.99, exactly matching the monthly price of the recurring 1.65M tier.

After the 1.1M-point project, that combination leaves 560,000 add-on points. Those add-on points remain usable for one year, while unused subscription points reset. The recurring $49.99 plan would instead leave 550,000 monthly points that disappear at the reset.

This is a timing arbitrage, not a permanent discount. If the operator needs about 1.1M every month, managing add-ons becomes needless work and the 1.65M recurring tier is cleaner. For one exceptional month, the smaller subscription plus longer-lived add-on preserves more future value.

A mid-market CTO with stable 2.4M-point demand

A measured workload of 2.4M points a month belongs on the 3.3M tier at $99.99. It leaves 900,000 points of headroom for variance. The 1.65M tier is predictably too small, and repeated add-on purchases would turn a stable budget into a manual process.

The 8.25M plan is not justified. Its cost per included point is essentially the same, so the CTO gains no volume discount by overbuying. The extra capacity only creates more expiring inventory.

Governance is the decisive purchasing question. If several people share one eligible subscription, any member can consume the common balance. A monthly owner, an approved-workload policy, and a Points History review are part of the plan cost. If the organization requires enforceable per-user limits, a direct team product can be the better buy even when its sticker is higher.

What Poe Free really includes

Poe Free is enough for casual chat when a daily reset does not interrupt anything important. The current Poe FAQ confirms that non-paying users receive points daily, those points reset after 24 hours, and they do not roll over.

The public pricing page does not promise a fixed free-point amount. That omission matters. A durable budget cannot be built around a number Poe does not publish. Treat the balance shown in your account as the current allowance and assume it can change.

Free users can allocate their daily points across bots whose rates fit the balance. Every bot displays its own approximate cost, and longer context or more expensive generation consumes more. The free tier is therefore an evaluation lane, not a guaranteed message bundle.

Who never needs to pay? A casual user who asks a few questions, can switch to lower-cost bots, and is willing to wait for the next 24-hour grant can remain free while the daily allowance continues to clear that work. The same is true for a founder using Poe only to compare occasional outputs before doing the primary work elsewhere.

Free stops being viable in three situations:

• A daily reset delays paid work.
• The models or context required cost more points than the available grant supports.
• Software needs to call paid models through the Poe API.

The API boundary is explicit. Paid-model API use requires an active subscription or sufficient add-on points. Free chat access does not create a free production API budget.

Hidden Poe costs

Poe's biggest hidden cost is not a fee. It is expired capacity. The sticker looks flat, but the value depends on when points arrive, when they reset, which bot consumes them, and whether one person can drain a shared pool.

Subscription points expire; add-on points last longer

Unused subscription points do not roll over unless a plan explicitly says otherwise. Daily points expire at the 24-hour reset. Monthly points expire with the monthly grant.

Add-on points behave differently. Poe's purchases FAQ says they are usable for one year, but they are non-refundable, non-transferable, and not redeemable for cash. That makes them useful for an irregular spike and poor for money you may never spend.

Annual billing locks the forecast

The displayed 17% saving assumes the plan remains right for a full year. Because annual billing breaks even in month 11 and charges are generally non-refundable, a buyer who stops after ten months would have spent less on monthly billing.

The risk is larger than cancellation. Point costs and included allowances have changed over time. The sticker can stay fixed while the amount of work it funds changes.

Buy annual only after observing a stable workload and accepting the current point mechanism. If a vendor changes the allowance, recalculate cost per consumed point instead of anchoring to the familiar monthly price.

The purchase platform can lock management

Poe says the website offers the most diverse subscription options. It also says upgrades must be managed on the platform where the active subscription was purchased. A subscription bought through Apple, Google, or the web stays tied to that channel until it is cancelled and replaced.

For the widest plan menu, buy on the web. This also avoids discovering during an urgent upgrade that the desired option is only available on another platform.

Prices are displayed in US dollars, but taxes and currency conversion can change the final charge. App-store refund rules also differ from the web route.

Refunds are an exception, not a planning tool

Poe allows cancellation at any time, effective at the end of the billing cycle. Access continues through the paid period. Fees and final charges are generally non-refundable except under the formal refund policy, applicable law, or a determination by Quora.

Web subscribers request eligible refunds through Poe support. Apple and Google subscribers use the respective store's policy. A business should therefore treat an annual purchase as committed spend, not money that can be reclaimed when priorities change.

Shared subscriptions have a common-pool risk

Eligible shared subscriptions do not provide per-member point controls. Poe says each member can individually use up to 100% of the subscriber's points.

That is not seat-based budgeting. A mid-market team cannot assume a fair split or divide the price by headcount. One large media job or agent run can consume the pool before other members start work.

If a shared subscription is used for business, assign an owner, define which workloads are allowed, and review aggregate usage. When predictable per-user governance matters, a direct team plan may be worth more than Poe's lower sticker.

Poe API pricing is a separate operating decision

Poe's external application guide says paid-model API calls require either an active subscription or sufficient add-on points. The API uses the same point economy; it does not turn a consumer subscription into unmetered infrastructure.

The important boundary is control. The owner of a Poe API key makes the account's entire point balance available to any bot queried with that key. A forgotten script, retry loop, or poorly bounded agent can therefore consume the same balance used for interactive work.

Set a per-message budget where Poe exposes it, review Points History, and separate experimental automation from business-critical capacity. A developer should know the expected point cost of a call before connecting a loop.

There is no single Poe token rate. Each bot exposes its own Rates view, and cost can depend on input, output, context, or the generated asset. The subscription table tells you how many points you own, not how many accepted outcomes those points will buy.

For an API-first workload, compare Poe with direct model pricing rather than with consumer app subscriptions. The DeepSeek pricing breakdown shows how a direct token bill behaves, while the cheapest AI API comparison helps identify when one-provider pricing beats an aggregator.

Poe is attractive when one point balance across many bots reduces operational friction. A direct API is stronger when one model dominates volume, native controls matter, or the business needs a contract that fixes governance and billing around a single provider.

Poe versus ChatGPT, Claude, and Perplexity pricing

Poe's 660K plan, ChatGPT Plus, Claude Pro, and Perplexity Pro all sit at roughly $20 a month. The stickers match; the products do not. Poe sells metered multi-model access, while each direct subscription sells its provider's native app, tools, and limits.

ChatGPT Plus: $20 monthly, no annual option

ChatGPT Plus costs $20 a month, billed monthly, and OpenAI does not offer annual Plus billing. It provides broader model and tool access than ChatGPT Free, while API use remains separately billed.

Choose ChatGPT Plus when the native ChatGPT product is the recurring workplace and other model families are occasional. Choose Poe when switching among providers is the work and a shared point meter is acceptable.

Over twelve months, ChatGPT Plus costs $240 because there is no annual discount. That is $40.01 more than Poe's $199.99 annual 660K plan, but the comparison is only sticker-level. Usage limits and native features are different.

Claude Pro: $20 monthly or $200 annually

Claude Pro costs $20 a month or $200 paid up front for a year, displayed as $17 a month with the annual discount. The direct plan is the clear choice when Claude is the dominant model and Claude's own product is where projects live.

Poe's annual 660K plan is one cent cheaper on the sticker. That difference is meaningless. The decision is whether a shared multi-model point budget creates more value than direct Claude access.

Perplexity Pro: $20 monthly or $200 annually

Perplexity Pro costs $20 a month or $200 a year. It is the direct alternative when cited research and search are the recurring outcome rather than general multi-model chat.

Poe and Perplexity solve different buying jobs. Poe is a general model marketplace with a point budget. Perplexity is built around sourced answers and research workflows. The full Perplexity pricing analysis covers its plan economics and usage costs.

The three-subscription comparison

Buying ChatGPT Plus, Claude Pro, and Perplexity Pro at their cheapest published annual path costs $640 a year: $240 for ChatGPT, $200 for Claude, and $200 for Perplexity. Poe's 660K monthly pool costs $199.99 on annual billing, a sticker difference of $440.01 a year.

That $440.01 is not automatic savings. It is the maximum budget released if Poe can replace all three direct products for the work you do. Poe does not give three independent native subscriptions, and its shared 660K points can run out.

For a founder who uses each model lightly, Poe can collapse three bills into one. For a research lead living in Perplexity, a developer living in ChatGPT, or a writer whose entire knowledge base sits in Claude, the direct product may repay its $20 through workflow depth alone.

The decision rule is simple: choose Poe for breadth, choose a direct subscription for depth, and pay for both only when each maps to a recurring paid outcome.

Price history and the current consequence

The current $19.99 plan includes 660,000 monthly points. A secondary pricing tracker, UsagePricing, captured the same sticker with 1 million monthly points before 28 July 2026, then recorded the move to 660,000. Poe's current page confirms 660K but does not publish that historical change as an announcement.

Treat the history as secondary evidence. The present price and allowance are first-party facts; the before-and-after date comes from the tracker.

The business consequence is still concrete. Moving from 1 million to 660,000 points is a 34% allowance reduction. At the unchanged $19.99 price, cost per included million points moved from $19.99 to $30.29, a 51.5% increase.

A workflow budgeted around the former allowance is now short 340,000 points a month. At Poe's advertised $30-per-million add-on rate, that gap represents $10.20 of point value before purchase-package sizing.

This is why sticker-price monitoring is not enough. The budget line did not move, but the amount of research, generation, or automation it can fund did. Operators should track cost per accepted result and points consumed, not merely whether the subscription still says $19.99.

The practical response is not to panic-upgrade. Check whether the lost 340K was ever used. A subscriber who historically consumed 400K remains inside the new allowance. A subscriber who depended on 900K now needs to reduce context, use lower-cost bots, buy add-ons, or select a larger pool.

Frequently asked questions

How much does a Poe subscription cost?

Poe has five paid options from $4.99 to $249.99 monthly. Annual prices run from $49.99 to $2,499.99 and are displayed at about 17% below twelve monthly payments.

Is Poe AI completely free?

Poe has non-paying access with a daily point grant. The points reset after 24 hours and do not roll over. Poe's public pricing and FAQ pages do not promise a fixed current grant, so Free is best for casual use that can tolerate the reset.

Does Poe offer a student discount?

Poe does not publish a standing student discount on its live pricing or purchases pages as of 17 August 2026. Students whose usage fits the daily grant can stay on Free; otherwise the same monthly and annual rates apply.

What is Poe's refund policy?

Poe fees and final charges are generally non-refundable, subject to its formal refund policy, applicable law, or a decision by Quora. Web purchases go through Poe support when eligible; Apple and Google purchases follow the relevant store's process.

Did Poe pricing change in 2026?

The current vendor page lists 660K monthly points at $19.99. A secondary tracker captured 1 million points at the same price before 28 July 2026, implying 34% less allowance and 51.5% higher cost per included point. Poe has not published that history as an announcement.

Does a Poe subscription include API access?

Poe lets paid-model API calls draw from an active subscription or add-on points. Those calls consume the account's point balance. A free account without sufficient add-on points cannot use paid models through the API.

The Monday move: audit the shape of usage

Do not choose from the pricing ladder on instinct. On Monday, turn one billing cycle of point data into a reversible plan decision.

1. 1

   Open Points History

   In Poe Settings, open Points History. Record the total points used, the busiest day's points, and the two or three bots responsible for most consumption. The busiest day decides daily versus pooled billing; the total decides pool size.

2. 2

   Classify steady or bursty

   If every representative day stays below 10K, use the $4.99 plan. If normal work crosses 10K on any day but the cycle stays below 660K, use $19.99. Do not average away the day that blocks delivery.

3. 3

   Separate spikes from baseline

   When usage above 660K comes from a one-time project, compare a $30-per-million add-on with the $49.99 recurring tier. Add-ons last one year; monthly allowance does not. Upgrade only when the higher baseline repeats.

4. 4

   Commit annually last

   Buy one monthly cycle first. Switch to annual only when the plan survived the busiest work and you expect at least 11 active months in the next year. Recalculate after any allowance or model-rate change.

This turns Poe from a vague AI subscription into an operating budget. The goal is not the largest point balance. It is the smallest recurring spend that never blocks a paid outcome.

Get the AI tools map for business owners

Match each AI subscription to the work, budget, and usage pattern it fits. Free to subscribers.