How does monday sidekick pricing work? The real credit numbers, as of 28 August 2026

If you have been trying to work out what monday sidekick will cost your team, you are not missing a page. monday's sidekick product page never uses the word credit. The pricing page shows a credit allowance in three of the four paid plan cards, and Custom AI credits on Enterprise. Four support articles carry monday's credit tables, and they do not all say the same thing.

There is a common assumption behind that: nobody can quote a number because monday does not own the models behind its AI, so cost must move with whichever model runs. For sidekick, monday's own documentation removes that variable.

In monday's AI models and credits article, a table has the columns Feature, Model select? and Credit use, and the monday sidekick row says No. Immediately above the table monday writes: "Other features currently have a credit cost regardless of the model." So for sidekick, the model is not a variable.

Everything below is quoted or calculated from monday's own surfaces, checked on 28 August 2026. monday's pricing-model article prints "Credit usage is subject to change", and monday edited two of these articles in the three days before that check, so the date is part of the fact.

There is no single switch-over date. monday documents three events, on three pages.

6 May 2026. monday's pricing-model article opens with its scope: it applies to customers who joined the monday Work Platform on or after 6 May 2026 and were offered to purchase seats and AI credits. It applies to the monday AI work platform only, not to monday CRM, monday dev or monday service.

8 June 2026. In "AI Permissions and Governance": "The Agent directory is currently in gradual release. Credit consumption for monday agents starts on June 8, 2026."

27 July 2026. This one appears on exactly one monday page. In "Get started with monday sidekick", under a section headed "Usage and limitations": "Starting July 27 2026, sidekick will consume credits." A month later, monday's sentence is still in the future tense.

The section heading reads "Usage and limitations" while its anchor link still ends in pricing. And monday's public changelog carries no entry about sidekick and credits between 31 May 2026 and 3 August 2026, and no entry at all between 15 July and 3 August. That window contains both the 8 June agents date and the 27 July sidekick date. Pages beyond the first were not readable when this was checked, so that bound is deliberate. A buyer looking for the announcement will not find one.

monday publishes the cash price of a credit in one place, the AI Feature Catalog. Under "Yearly plans: Cost per 1 credit" the US dollar cell reads "$0.01". Under "Monthly plans: Cost per 1 credit", with a separate line reading "(25% higher than Yearly plans)", it reads "$0.0125".

Direction matters and is easy to reverse. Monthly is 25 percent higher than yearly, which is the same as saying yearly is 20 percent lower than monthly.

If you buy in another currency, monday prints the price in ten and its "(25% higher than Yearly plans)" label is exact for 6 of the 10. GBP is about 28.6 percent higher, CAD about 27.3 percent, AUD about 23.1 percent and ILS about 23.3 percent, almost certainly rounding. Only the US dollar cells carry a currency symbol.

monday's pricing-model FAQ says that for customers signing up from 6 May 2026, "AI credits must be purchased alongside seats. They are not included for free." The same article gives the minimum monthly amount as 1,000 credits for Basic, 2,000 for Standard and 3,000 for Pro, with Enterprise handled by sales. At purchase you can pick more: Standard offers 2,000, 4,000 or 8,000 a month, and Pro offers 3,000, 4,000, 8,000 or 20,000. monday's pricing page prints the same 1,000, 2,000 and 3,000 in its plan cards.

The same plan cards print the cash equivalent: 1,000 credits at $10 a month, 2,000 at $20 and 3,000 at $30. The largest Pro bucket of 20,000 comes to $200 at the same $0.01, which is our arithmetic on monday's figures. monday notes that some customers receive discounted introductory pricing, so a published figure is a list price.

monday puts every sidekick message in one of four complexity bands. The AI Feature Catalog gives the sidekick row as simple at roughly 10 to 30 credits, intermediate at roughly 30 to 80, complex at roughly 80 to 150, and extra complex at roughly 150 or more. monday's AI models and credits article prints the same four under a column headed "Est. credits per message".

monday publishes a second set of four bands for monday agents, with the same labels and different numbers. The two sets appear together in monday's articles and are easy to confuse. In monday's models and credits article the agents figures sit under "Est. credits per task" and sidekick's under "Est. credits per message". Check which row you are reading. Every band in this post is the sidekick row. A third set, for monday vibe, is different again.

What moves a message between bands is not the model. monday's sentence: "Consumption depends on the depth of the request, how much board context is read, and which tools are used, such as search, updates, charts, or asset generation." monday also says consumption is based on the complexity of each interaction, from the first message.

A simple message at 10 to 30 credits is $0.10 to $0.30 at the yearly price, and an extra complex one from 150 credits is from $1.50. That arithmetic is ours, on monday's price and monday's bands.

Under "Sidekick examples in action", monday works out four recurring habits and prints a monthly credit figure for each.

Daily board status checks: "Credits potential spend: ~770 per month (22 workdays)".

Ticket triage and routing: "Credits potential spend: ~19,800 per month (15 tickets per day)".

Building charts and reports: "Credits potential spend: ~660 per month (10 reports)".

Bulk board updates: "Credits potential spend: ~1,256 per month (4 batch jobs); heavy one-off rebuilds can run significantly higher".

Read the bracketed scope on each. At the yearly credit price those are about $7.70, about $198, about $6.60 and about $12.56 a month, and on a monthly plan at $0.0125 about $9.63, about $247.50, about $8.25 and about $15.70. The conversion is ours; the credit figures are monday's.

One daily status-check habit on one board, at about 770 credits, is 77 percent of a Basic account's monthly 1,000. The ticket triage example, at about 19,800, is 99 percent of the largest Pro bucket of 20,000 and 6.6 times the Pro minimum of 3,000. Those comparisons hold only if you buy the minimum bucket.

monday labels all of this: "Estimates are directional and may vary based on board size, conversation length, and the tools invoked", and "Credit amounts, ranges, and examples shown in this article are intended as general guidance".

Mostly yes, and in two places no, one hard and one soft.

Three of monday's four worked examples divide out inside its own band table. 770 credits over 22 workdays is 35 a message, inside the intermediate band. 660 over 10 reports is 66, also intermediate. 1,256 over 4 batch jobs is 314, above the extra complex floor. The fourth, ticket triage, prints a monthly figure without a per-message count. The rate card holds together.

The gap is between monday's marketing surface and monday's support surface. On the pricing page monday says a typical 3-person team runs on roughly 800 to 1,200 credits a month, and that "1,000 credits can cover roughly 50 resume screenings, 5 hours of meeting summaries, and hundreds of automated workflow updates".

monday publishes a rate for one of those three. In three support articles, against the row label AI Notetaker, the cell reads "120 credits per meeting hour". Five meeting hours at 120 credits is 600 credits, our multiplication of two monday figures. That is 60 percent of the 1,000, leaving 400 for the resume screenings and the workflow updates combined.

We stop there deliberately. monday has no sidekick credit rate for a resume screening or for an automated workflow update, so the remaining 400 cannot be checked against any published sidekick rate. Dividing 400 by 50 would produce a number nobody has published.

The second conflict is softer. monday's own "Bulk board updates" example, at about 1,256 credits a month, exceeds the top of monday's own typical range of 1,200 by 56 credits, about 4.7 percent, on its own. The daily status-check example at about 770 is 64 percent to 96 percent of that range. All four together total 22,486 credits, about 18.7 times the top of the range.

None of that means the rate card is broken. It means monday's pricing page describes a lighter pattern of use than monday's support articles do.

Four places, all live on 28 August 2026. Two of them change what you should plan for, and two are smaller. Naming the surface rather than the company is the only honest way to report them.

What happens at the limit. Three monday articles say notifications fire at 80% and 100% of usage, and that "some of your AI capabilities will continue running for a short period after the limit is reached", giving you time to top up before usage is paused. The AI models and credits article says instead that "you will need to add more credits to your plan in order to continue using paid AI capabilities". Both are current, and their last content edits are 34 seconds apart, so recency cannot settle it. Plan for the stricter one.

How you top up. Three monday articles point you at a Buy more credits button in the administration section. The newest-edited of the four credit articles, the AI Feature Catalog, says instead: "Contact our support team or your Account Team to upgrade to the next AI credit bucket." Treat them as alternatives, not a sequence.

Two smaller ones. monday's articles call the admin screen Credits Usage in one place and the Credits tab in another, and both sit under Administration, then AI governance. And the sidekick product page says "monday sidekick is available on all monday.com workspaces" with "no setup required", while the pricing-model article says "AI features are available for all paying accounts, and are not available for free accounts." The plan cards settle that one: "Sidekick AI assistant" is listed on all four paid tiers under "Powered by AI credits:", so read it as a paid-plan feature from Basic upwards.

No credit-to-token mapping. monday names four model providers and explains, qualitatively, that a more capable model costs more credits. It publishes no conversion between a credit and any token or model usage. The precise version matters: monday does publish per-model credit bands for monday vibe, naming Gemini Flash among others, so it is not silent on models and credits. For sidekick it says the model does not change the cost anyway. Its AI FAQs answer directly: "Currently, monday AI does not support bringing your own model or using custom API keys."

That answer is scoped to monday's own AI features. Bring Your Own Agent runs on your own provider account, and monday's setup instructions include pasting your own API key for several providers. monday's table says reasoning on the external platform consumes "Third-party provider credits (for example, Anthropic tokens), not monday AI credits". monday labels Bring Your Own Agent "currently in Beta", and adds that it "is currently available at no additional cost. This may change in the future." That split does not reach sidekick, which has no model choice and no third-party credit line anywhere in monday's documentation.

Across the five articles that define the credit model, monday does not say what happens to unused credits at the end of a billing cycle. monday has not said credits expire, and neither will we.

No published cost for sidekick Routines, and this is the one to watch. monday's sidekick product page describes Routines as "scheduled AI actions in monday sidekick that run automatically on a set cadence", configured from the Centerkick sidebar and running "without manual prompts". Sidekick is metered per message, and Routines run sidekick actions on a schedule with nobody present. As of 28 August 2026 the help centre does not document Routines at all, including their cost. That describes monday's documentation, not the product: it is not evidence they are free, or that they are uncapped.

monday's governance article documents usage limits at "three levels: account-wide, per AI capability, and per user". Each feature limit is labelled Hard or Soft, which sets the enforcement at the limit, and you can set a default user limit plus higher policies for power users.

On reporting, the credits view shows credits used in the current billing cycle and your total balance, with a chart projecting usage at the current rate, +10% and +20%. You can review consumption per feature and per user, filter it, and download the data. Accounts on trial AI credits do not see the projection widget. monday's article on available plan types prints a one-time trial of 6,000 AI credits on Basic, Standard and Pro, and 12,000 on Enterprise.

monday publishes exactly one flat per-action rate for the AI work platform, "8 credits per action." for AI blocks, with all actions on the same item in a 24-hour period counted once. monday has committed to 30 days' notice before changing a credit rate, but only in its article on monday Smart Call, which is a monday CRM feature: you will be notified "at least 30 days in advance", and the change "will only apply to new billing cycles". No equivalent commitment appears in the articles that govern sidekick. monday's pricing-model article says instead "Credit usage is subject to change".

If your account joined monday on or after 6 May 2026, you can build a genuine forecast today. Take monday's four worked examples, match them to your team's habits, multiply by the credit price for your billing cycle, and set per-feature and per-user limits. Plenty of teams can do that themselves, and if that is where you land, take the numbers above and go.

If your account joined before 6 May 2026, monday's pricing-model article is scoped to customers who joined on or after that date, so check your own agreement, and check monday's article on available plan types for the AI work platform.