Comparison
PM Tool AI Credits Explained 2026: What Asana, ClickUp, monday.com, Notion and Wrike Actually Charge For
Five vendors now meter AI in five incompatible units — requests, pooled credits, per-account credits, and per-member allowances that expire. Here are the verified 2026 rates, the cost of the same workload in each, and why 'AI included' is the most expensive sentence in the category.
AsanaClickUpmonday.comNotionWrikeSmartsheet
Every project management vendor added an AI meter in 2025 and 2026. No two meters count the same thing. One charges per request, one per credit where credits are pooled per workspace, one per credit that belongs to the account rather than the seat, one in two per-member allowances that do not convert into each other, and one advertises an “AI Elite starter pack” with no published rate at all.
That is not a detail for procurement to sort out later. It is the difference between an AI line item that costs $240 a year and one that costs $4,800 for the same work — and because the units are not interchangeable, the vendor’s own comparison table will not tell you.
This page converts the published 2026 rates into one comparable form, prices the same workload on each vendor, and answers the question that actually decides a purchase: when does metered AI cost more than the seats it sits on?
TL;DR
- The units are incompatible and the vendors know it. Asana charges $0.50 per AI request prepaid ($0.60 pay-as-you-go). ClickUp sells top-ups at $10 per 10,000 credits. Notion sells $10 per 1,000 credits monthly, $13 per 1,000 annual. monday.com includes credits in the plan (1,000 / 2,000 / 3,000 by tier) and sells more as an add-on whose per-credit rate is not on its pricing page. A “credit” in these five products is not the same quantity.
- The cheapest published AI unit in the category is ClickUp’s $10 per 10,000 credits. The most expensive is Notion’s annual pool at $13 per 1,000. That is a 130× difference in nominal unit price — which tells you almost nothing until you know what a credit buys, and vendors do not publish a credits-per-action table. Treat unit price as a tiebreaker, never as the decision.
- The dangerous line is not the overage price, it is the allowance that expires. Asana’s paid plans include “5 requests per user per month, up to 250 requests per account” — a per-account ceiling, not a per-seat allowance, so a 100-person company gets the same 250 AI requests as a 10-person one. Notion’s per-member credits reset monthly and “unused credits don’t roll over.”
- Asana is the only vendor here that publishes a hard per-account cap on included AI. At $0.50 per request, a mid-size team that uses AI in one workflow will exceed 250 requests in the first week and be buying packs from then on. Model that as a recurring cost, not a spike.
- Two vendors do not publish a rate at all. monday.com shows what included credits are worth (“3,000 credits = $30” on Pro) and says “Add credits anytime,” but the purchase rate for more credits is not a published number. Wrike’s pricing page lists an “AI Elite” starter pack on Business without a per-action price. Both require a quote — do not estimate them, and do not accept a competitor’s estimate either.
- If you are under 10 people: the AI meter is almost never your largest cost. Seat price dominates by an order of magnitude. Buy for the work, not for the AI allowance, and pick the vendor whose AI is included rather than metered if your usage is light and unpredictable.
The five meters, side by side
All figures below come from each vendor’s own pricing page, read on 2026-11-07. Per-user prices are billed annually unless a monthly figure is given.
| Vendor | What the meter is called | What a seat includes | How you buy more | Published rate |
|---|---|---|---|---|
| Asana | AI requests + AI Studio credits | 5 AI requests / user / mo, capped at 250 / account; AI Studio 50,000 / 75,000 / 200,000 credits per billing account / mo (Starter / Advanced / Enterprise) | Prepaid packs or on-demand | $0.50 / request prepaid, $0.60 / request on demand |
| ClickUp | AI credits (pooled) | Free: 500 / workspace. Core: 1,000 / user / mo. Business: 1,500 / user / mo. Enterprise: 3,000 / user / mo. Max: 10,000 / user / mo. All pooled at workspace level | Top-up on any plan | $10 per 10,000 credits |
| monday.com | AI credits (per account) | Basic 1,000 / mo, Standard 2,000 / mo, Pro 3,000 / mo (shown against a 10-seat default order) | Add-on, “Add credits anytime” | Not published on the pricing page |
| Notion | Standard credits + Select credits + Notion credits | Business: 1,500 Standard + 500 Select / member / mo, neither convertible nor transferable | Notion credits (Business and Enterprise only) | $10 / 1,000 monthly, $13 / 1,000 annual |
| Wrike | AI Essentials (Team), AI Elite actions (Business) | Team includes “AI Essentials”; Business includes “a starter pack of AI Elite actions” | Not specified on the pricing page | Not published |
| Smartsheet | No credit currency | Smart Assist, Smart Columns, “Generate formulas”, AI dashboard builder — listed as included from Pro up | n/a | No published metering |
Verified against Asana pricing, ClickUp pricing, monday.com pricing, Notion pricing with Notion Help — What are Notion credits?, Wrike pricing and Smartsheet pricing on 2026-11-07.
Three structural differences jump out before any arithmetic.
Per user, per account, or per workspace. ClickUp pools by workspace but allocates by user (1,500 per user on Business). Notion allocates by member and does not let members share. monday.com’s published allowance is per account (1,000–3,000 total), which makes it the most constrained of the four at the top of the range: a 50-seat Pro customer gets the same 3,000 credits as a 10-seat Pro customer. Asana is per account for requests (250 ceiling) and per billing account for AI Studio credits.
Convertible or not. Notion’s Standard and Select credits are explicitly non-transferable — “you can’t combine them, trade one kind for another, or move them between members.” ClickUp’s are pooled, which means unused capacity in one team balances a heavy user in another. A pooled allowance is worth more than the same number of non-pooled credits, and no vendor’s headline number accounts for that.
Expiring or prepaid. Asana’s prepaid AI requests are “yours for your full contract terms with no monthly resets,” while its on-demand rate costs 20% more. Notion’s monthly credits reset and “unused credits don’t carry over,” while annual credits are prepaid and usable all year but also do not carry into the next subscription term. Expiry behaviour decides whether “buy in bulk” is a discount or a bet.
The same workload, priced five ways
Here is the exercise that makes the comparison usable. Define one workload and price it on each vendor. The assumptions are stated so you can substitute your own numbers.
The workload: a 20-person team, one AI workflow — an intake agent that reads a submitted request, summarises it, proposes an owner and writes three subtasks. Say the workflow runs 40 times a week, so 160 runs a month, and each run is one billable unit in the vendor’s own terms.
| Vendor | Billable units / mo | Included in 20 seats | Units to purchase | Published cost of those units |
|---|---|---|---|---|
| Asana (Advanced) | 160 requests | 250 / account ceiling (min of 5×20=100 and 250 → 100) | 60 | $30 / mo at $0.50 prepaid |
| ClickUp (Business) | 160 runs, pooled | 30,000 credits / mo (20 × 1,500) | 0, if a run costs under 187 credits | $0 |
| monday.com (Pro) | 160 runs | 3,000 credits / mo (per account) | Only if a run exceeds ~18 credits | Quoted — rate not published |
| Notion (Business) | 160 runs | 40,000 credits / mo (20 × 2,000 combined) | 0, if a run costs under 250 credits | $0 |
| Wrike (Business) | 160 runs | “Starter pack of AI Elite actions” | Unknown | Quoted — rate not published |
The table’s most important column is the one that says “if a run costs under X credits.” Notion and ClickUp both have allowances generous enough to absorb this workload, and both refuse to tell you how many credits a run costs. That is the honest state of 2026 AI pricing: the vendor publishes the size of the bucket and hides the size of the ladle. The only people who know are customers who measured it, and their numbers are in the next section.
Only Asana can be priced outright, because it charges per request — a unit tied to a deliverable rather than a token or credit. That is a genuine advantage in one narrow sense: you can predict Asana’s AI bill from your task volume before you use the product. It is also why Asana’s bill arrives sooner. At 160 runs a month against a 100-request allowance, you are buying 60 requests in month one.
The Asana arithmetic, because it is the only one you can finish
Asana’s included allowance is the constraint on this page worth designing around: “5 requests per user per month, up to 250 requests per account.” That ceiling is per account, not per seat, which inverts the usual intuition.
| Team size | 5 requests × seats | Actual ceiling | Requests per person |
|---|---|---|---|
| 10 people | 50 | 50 | 5 |
| 30 people | 150 | 150 | 5 |
| 50 people | 250 | 250 | 5 |
| 100 people | 500 → capped | 250 | 2.5 |
| 500 people | 2,500 → capped | 250 | 0.5 |
At 50 seats the two rules meet; above that, every additional seat reduces each person’s share of included AI. A 200-person Asana customer has 1.25 AI requests per person per month included in the platform price. If your company’s plan is “everyone uses the AI assistant occasionally,” that is not an allowance, it is a sample.
At $0.50 prepaid, 1,000 additional requests a month is $500; at $0.60 on-demand it is $600. For a 200-person team running AI in three workflows, the AI line item can plausibly exceed the per-seat difference between Asana and a cheaper tool. That is the moment “we chose Asana for the AI” turns into “we chose Asana and now pay twice.”
Note what this does not say: it does not say Asana’s AI is worse. Request-based pricing is predictable and simple, and Asana’s AI Studio credits (50,000–200,000 per billing account) are a separate, larger pool. The claim here is only about the size of the included allowance and the ceiling that stops scaling with headcount.
Where the allowances are actually generous
Two vendors’ included allowances are large enough that most teams will never see an overage, and both share a reason: they allocate per user and pool.
ClickUp. Business includes 1,500 credits per user per month, pooled. For 20 seats that is 30,000 credits a month before you buy anything. The top-up rate when you do is $10 per 10,000 credits — the cheapest published AI unit in the category by a wide margin. Whatever a credit represents, $0.001 per credit means a 30,000-credit allowance is worth $30 at the margin, and the marginal cost of a heavy AI month is small relative to the seat price of $17 per user.
Notion. Business includes 2,000 credits per member per month (1,500 Standard + 500 Select), and at $10 per 1,000 the replacement value of that allowance is $20 per member per month — equal to the entire seat price. That framing cuts both ways: the allowance is worth exactly one seat’s price, which makes it the single most valuable thing in the Business plan.
monday.com is the odd case: 1,000 / 2,000 / 3,000 credits per tier, shown against a 10-seat default. For a 10-seat Pro account that is 300 credits per seat; for a 50-seat Pro account, 60 credits per seat. The allowance is flat while the seats grow, so the per-seat value of monday’s included AI falls as your team grows — the same structural criticism as Asana’s 250 ceiling, in a different shape.
What breaks in production
Every one of these meters has the same failure: an automated workflow keeps spending after the humans have stopped watching. The published rates do not describe this, but the customers do, and the pattern is identical across vendors.
The runaway automation. A team reports running through a month of AI Studio credits in hours: “It is very easy to blow through a month’s worth of credits in hours if you don’t test and prepare accordingly.” Their fix is method, not tooling — benchmark representative use cases in a private project, establish credits-per-action for each, multiply by expected frequency, then compare against the budget. (Asana Forum — “Manage AI Use Credit Expectations”)
The builder that looked like a helper. In the same thread: “The screen to start a new workflow or automation also pushes them towards using AI Studio. Our users have been defaulting to the easier method and we ran out of credits in 2 weeks… The two users that burned through all the AI Studio credits had no idea that the workflow would use AI, they just assumed AI was helping build it out.” That is a UX failure with a bill attached, and it exists on every vendor’s AI builder, not just Asana’s.
The prompt that spends more than it needs to. Another reply in the same thread documents the mechanism: an AI step asked to create a task from a form submission “would often stretch its logic… The extra tasks were annoying to consolidate, but it also was burning a lot of tokens trying to parse the plurals.” Credit spend is sensitive to prompt wording — which means credit budgets are an editorial problem as much as a finance one, and prompts deserve version control.
The visibility gap. Users cannot reliably tell whether an action cost a credit before it happens: “nothing tells me which side (free or paying) before the result, or even after…” (Asana Forum — “Dash: clarifying if a message costs a Request”). A forum leader’s summary of the problem is the best one-line description of 2026 AI pricing: “The most confusing thing ever is these AI credits. When you have to create lengthy explanations for users to understand, or use AI to explain AI, then it’s time to rethink the model and dumb it down.” (Asana Forum — “Credit awareness for builders”)
The missing stop button. The same thread contains what a finance owner would ask for and does not get: “The best outcome would be to have the ability to set limits that will ensure that process interruption is not going to occur because budget for credits is exceeded… There should be a way to limit new AI rules until credits are added to the account.”
A caveat about this evidence: these are Asana forum threads, dated 2026, on one vendor’s product. We cite them on a five-vendor page because the mechanics — a builder defaulting to AI, prompts affecting token burn, no pre-action cost signal — are properties of metered AI in general, not of Asana’s implementation. When another vendor’s community produces equivalent posts, the same design rules apply.
How to budget metered AI without a rate card
Since no vendor here publishes credits-per-action, the budget has to be built from measurement. This is a five-step method, drawn from what paying teams actually did rather than from a vendor template.
- Pick the five AI actions your team will run most. Not “AI,” specific actions: summarise this request, draft this status update, generate these subtasks, translate this page, answer this question about last quarter.
- Measure each one in a private project. Run it ten times with realistic inputs, record the credits consumed, and take the average. Use realistic messy inputs, because prompt-shaped inputs understate consumption.
- Multiply by frequency, not by seats. A 20-person team where two people use an action daily consumes more than a 20-person team where everyone uses it monthly.
- Add the surprise budget. Set 20–30% aside for the workflows nobody modelled — the ones a functional lead builds in an afternoon because the builder made it easy.
- Re-measure quarterly, and re-measure after any vendor repricing. Allowances in this category moved at least twice in 2026 across the vendors on this page.
Then apply the one rule that survives every vendor change: keep the AI workflow on a deterministic alternative. Every one of the tools here has a rules-based automation engine that does not consume AI credits, and in many workflows it is the better instrument anyway. A forum leader who has implemented this for clients puts it bluntly: helping clients “use regular automations instead of AI Studio automations in many instances. It’s always cheaper, often simpler, and sometimes better (deterministic instead of probabilistic).”
That is the single highest-value sentence on this page. The cheapest AI credit is the one you never spend because a rule could do the job.
Team-perspective judgement
Under 10 people, light AI use (a few summaries a week). Ignore the AI meter entirely and choose on seats and fit. A 6-person team on ClickUp Business has 9,000 pooled credits a month; on Notion Business, 12,000; on Asana Advanced, 30 included requests (6 × 5). Only the Asana number is small enough to notice, and even then the overage for a light month is a few dollars. Buy the tool your team will use.
Under 10 people, one AI workflow that runs daily. This is the case where the meter becomes real. Asana hits its 30-request included allowance in about a week and then costs $0.50 per run — a 40-run-a-week workflow is $80 a month on top of seats. ClickUp and Notion absorb it inside their allowances and cannot quote you a per-run cost. For a team this size, the pooled per-user allowance is worth more than the published overage rate.
10–40 people, AI in three or more workflows. Budget AI as a fourth line item next to seats, and expect it to be 10–30% of the software bill. Two structural checks before you sign: does the included allowance scale with seats (ClickUp and Notion: yes; monday.com: no; Asana requests: no, capped at 250), and is the allowance pooled (ClickUp: yes; Notion: no, per-member and non-transferable). A non-pooled, non-scaling allowance fails first.
40+ people, any vendor. Assume the included AI allowance is a sample, and negotiate. This is the size at which the published rate stops being the price: at 40+ seats you are in “ask for a quote on credit volume” territory with all four metered vendors, and the list price is a starting point. Get the credits-per-action behaviour in writing for your top three workflows — a vendor that cannot answer that question has not priced their own product.
Any size, deciding between two vendors you like equally. Compare the included allowance at your headcount, not at the default on the pricing page. monday.com’s page quotes credits against a 10-seat order, and Asana’s 250-request ceiling looks generous until you divide it by 50. The number that matters is per-person included AI at your actual seat count: ClickUp at 1,500 per user does not change with headcount; Notion at 2,000 per member does not change either; Asana and monday.com both fall as you grow.
Any size, in a regulated or audited environment. AI credits are the least of your problems, but note one asymmetry: Asana’s prepaid packs are “yours for your full contract terms with no monthly resets,” which is the only genuinely budget-friendly structure here for finance teams. Everything else expires monthly, which means an unspent allowance is a write-off, not a reserve.
Related reading on this site
- The detailed single-vendor trackers: Notion pricing changes 2026, Notion AI updates 2026, ClickUp pricing plans explained, Asana new features and pricing, monday.com plans and pricing, and the AI pricing and feature tracker.
- Feature-level AI comparisons: Asana vs ClickUp vs monday AI features, ClickUp vs Notion AI features, AI project management tools 2026.
- Seat-level pricing comparisons: monday vs ClickUp vs Asana pricing, ClickUp vs monday vs Asana for enterprise, Asana vs ClickUp vs monday head-to-head.
- The other hidden cost in the category: who owns your PM data? Export and lock-in.
- Building the workflows that spend credits: ClickUp automations step-by-step, build an Asana workflow for agencies, reduce meeting overload with PM tools.
- The full field: best project management software 2026.
How this content was prepared
Transparency note: This comparison draws on three sources. (1) Vendor primary pricing documentation, read on 2026-11-07 and linked at each claim: the public pricing pages of Asana, ClickUp, monday.com, Notion, Wrike and Smartsheet, plus two Notion Help articles that define the three-credit model (“What are Notion credits?” and “Standard, Select & Notion credits: eligible features”). Every rate and allowance on this page — $0.50/$0.60 per request, $10 per 10,000 credits, 1,500/1,000/3,000/10,000 per-user credits, $10 and $13 per 1,000 Notion credits, the 250-request ceiling, 1,000/2,000/3,000 monday credits — is quoted from those pages. (2) Real-user evidence from vendor community forums, quoted inline with links: four Asana Forum threads from 2026 documenting runaway credit consumption, credit-visibility gaps, prompt-driven overspend and the absence of hard limits. We cite one vendor’s community on a multi-vendor page because the failure mechanics are general; where the evidence is vendor-specific, we say so. (3) Editorial judgement, expressed as the normalisation table, the 20-person workload model, the 250-ceiling arithmetic and the team-size recommendations.
We have not independently measured credits-per-action on any of these products, and we say so rather than substituting a number. No vendor publishes a credits-per-action table, which is why the “included allowance” column on this page carries a conditional (“if a run costs under X credits”) rather than a fabricated per-run cost. monday.com’s per-credit top-up rate and Wrike’s AI Elite action price are not published on their pricing pages; this page marks them as quoted and does not estimate them. Both vendors’ list prices are shown as published and nothing more.
Where we do make an assertion, it is this: at 40 seats or more, the included AI allowance on three of the five metered vendors stops scaling with your team, and the AI line item becomes a recurring cost that is invisible in the per-seat comparison every buyer uses. Ask for the credits-per-action number for your own top three workflows before you sign — a vendor that cannot provide it has not yet priced what it is selling you.
Verified against Asana pricing, ClickUp pricing, monday.com pricing, Notion pricing, Wrike pricing and Smartsheet pricing on 2026-11-07.