Devin Pricing 2026: Free, Pro, Max and Teams Explained (Windsurf Included)

Devin costs $0 on Free, $20 a month on Pro, $200 a month on Max, and $80 a month plus $40 per full seat on Teams, with Enterprise on a quote. The old Agent Compute Unit pricing is gone: a paid plan now buys a usage quota that refreshes daily and weekly, and anything past it is billed at API pricing. Prices read on Devin's pricing page on October 26, 2026.

One pricing page now covers two brands. Windsurf's editor is sold as Devin Desktop, and windsurf.com/pricing serves the same four plans, so a Windsurf buyer and a Devin buyer are reading the same ladder. The subscription covers cloud agent sessions, the Devin CLI and the desktop editor from one quota.

An AI coding agent is a tool that takes a task in plain language, then plans, edits, runs and tests code on its own, in your editor or in a cloud sandbox, and hands back a pull request. Toolradar data: the AI coding ranking we publish tracks 179 AI coding tools as of October 25, 2026, and Devin is the third most cross-shopped of them, listed as an alternative to 104 other tools, behind Cursor at 140 and GitHub Copilot at 129. For the broader category, see our best AI coding assistants and best AI agents rankings.

Devin plans at a glance

Plan	Price	Who it is for	What the plan lists
Free	$0	Trying the editor	Light quota to code with agents, limited model availability, unlimited inline edits and Tab completions
Pro	$20 a month	One developer using Devin most days	Increased quotas with frontier models from OpenAI, Claude, Gemini and SpaceXAI, full model availability, Devin Cloud agents, extra usage at API pricing
Max	$200 a month	One developer who keeps hitting the Pro quota	Everything in Pro with significantly higher quotas
Teams	$80 a month plus $40 per full seat	Teams up to 200 users	Everything in Pro, sharing and collaboration, centralized billing, admin analytics, priority support
Enterprise	Contact sales	Companies that need SSO and a private deployment	SAML/OIDC SSO, VPC deployment, admin controls, dedicated account team

Pro and Max are single-user plans. Cognition's self-serve plans documentation says they cannot be shared, so two developers on one login is a Teams account, not a Pro account.

How much does Devin cost for a solo developer?

A solo developer pays $240 a year on Pro, or $2,400 a year on Max, both billed monthly (our arithmetic, 12 times the printed monthly price). The pricing page prints no annual rate, so there is no discount to chase. Which of the two you need depends on how often the quota runs dry, and the pricing page's FAQ says most users never hit their limits.

Pro is the plan to start on. It adds the frontier models, free use of Cognition's own SWE-2 model, access to Devin Cloud (the sandboxed agent that runs without your laptop), DeepWiki, the Devin API, Ask Devin, and the Slack, Teams, Linear, Jira, GitHub, GitLab and Bitbucket integrations that Free lacks. Free, by contrast, has no cloud agent at all: it is the desktop editor with a light agent quota, unlimited inline edits and unlimited Tab completions, and the compare table marks premium models as unavailable.

The Pro weakness is the daily cap. A Pro quota refreshes on a daily and a weekly basis; a long afternoon of agent sessions can empty the day before the week is spent. The docs say the daily allowance is more than one seventh of the weekly, so working through a weekend does not cost you week-long capacity, but a single heavy day still stops at the daily line unless you buy extra usage.

Devin Max: who should pay ten times Pro?

Max removes the daily cap and keeps only a weekly allowance, which is the real reason to buy it. The pricing page describes Max as everything in Pro with significantly higher quotas and tags it as new. The self-serve documentation adds the detail that matters: Max has a weekly quota only, shared between Devin Cloud, the CLI and Devin Desktop, with no daily limit.

That makes Max the plan for someone who runs several cloud sessions in parallel for hours (both Pro and Max allow up to 10 concurrent sessions) rather than someone who uses the editor's agent a few times a day. If your pattern is bursty, say two heavy days a week and nothing in between, Max fits. If your pattern is steady and moderate, Pro plus the occasional top-up is cheaper, because extra usage is consumed at API pricing and purchased credits never expire.

The weakness is that Cognition publishes no token or session count for either quota. "Significantly higher" is the only sizing you get, and cost per message varies with the model, the task size and the reasoning required. You find out where the ceiling is by hitting it.

Devin Teams: the account minimum and the two seat types

A Teams account costs at least $80 a month, and a full developer seat costs $40 a month, with the seats counting toward that minimum. The pricing page prints the team plan and the seat as two separate lines. The documentation explains how the two numbers interact, and it is less expensive than the headline reads.

Every member holds one of two seat types:

• A full seat is a fixed monthly line at the per-seat price and includes its own daily and weekly quota equivalent to Pro plus Devin Desktop access.
• A flex seat is free, has no fixed monthly charge, draws entirely from the team's shared pool of on-demand credits, and does not include Devin Desktop.

The minimum can be met by any mix of full seats and credits. With zero full seats you pay the minimum and get all of it as shared credits, and with one full seat the other half of the minimum is charged as credits. From two full seats up the minimum is cleared and the bill is the seat price times the number of seats, which means a two-person team pays nothing for the "team plan" line at all. Here is the team cost at four sizes, from Cognition's own minimum table and our arithmetic:

Team shape	Monthly bill	What you get
0 full seats, flex users only	The minimum	All of it as shared on-demand credits
1 full seat	The minimum	One Pro-equivalent quota plus half the minimum in shared credits
2 full seats	The minimum	Two Pro-equivalent quotas, no included credits
5 full seats	$200	Five Pro-equivalent quotas, $2,400 a year
10 full seats plus 20 flex users	$400 plus credits you top up	Ten quotas, 20 occasional users on the shared pool

Teams also unlocks what Pro cannot: unlimited concurrent sessions, sharing and collaboration, a dedicated Slack Connect support channel and an optional onboarding call. Automations and Devin Review draw from the shared credit pool on Teams, not from seat quotas, and when the pool runs out Automations stop until you top up.

The cap moved this autumn. A snapshot of the pricing page from September 17, 2026 listed unlimited team members and unlimited flex seats; the live page now says up to 200 users, and the documentation sends larger teams to Enterprise.

Enterprise: the only plan still priced in ACUs

Enterprise has no public price. Cognition's enterprise billing documentation says Enterprise customers are billed in Agent Compute Units at the rate set in their order form, and that Enterprise ACUs are distinct from self-serve quota and credits. Admins can set per-organization and per-user ACU limits, which is the kind of control the self-serve plans do not offer.

The plan adds SAML/OIDC SSO, deployment in your own VPC, teamspace isolation, centralized management of several Devin organizations, custom git provider support, custom terms and a dedicated account team. If you are asking what Enterprise costs, you are reading the wrong plan: the buyer is a platform or security lead with a procurement process, not an individual.

What happened to ACU pricing?

Devin's self-serve plans dropped ACUs in spring 2026, and the entry price for a team fell from $500 a month to the current Teams minimum. A snapshot of the pricing page from March 22, 2026 still showed the old ladder. Core was pay as you go at $2.25 per ACU with a small starting minimum, and Team cost $500 a month with 250 ACUs included at $2.00 per ACU. The snapshot from April 24, 2026 shows the current Pro, Max and Teams plans instead, so the switch happened between those two dates.

The change is bigger than a price cut. Under ACUs you prepaid a normalized unit of compute and watched a balance drain per session. Under the quota model you prepay a plan, get a daily and weekly allowance, and only touch per-use pricing past it. For a solo developer the practical difference is predictability: the Pro price is the bill, unless you choose to buy more. For a five-person team the difference is $500 a month for 250 shared ACUs versus five Pro-equivalent quotas for less than half that (our arithmetic, from the team table above), with the old unit surviving only inside Enterprise contracts.

The Free plan is new too. The ACU-era page had no free tier; the cheapest way in was the Core minimum. Today's Free plan gives you the desktop editor and a light agent quota without a card.

Devin vs Claude Code vs Cursor at the same price points

At $20 a month, Devin Pro buys a cloud agent and a desktop editor, while Claude Pro buys Claude Code plus the Claude apps; Cursor has no plan at that price. At the top of the ladder all three vendors sell a tier at the Devin Max price, and they size it differently. Prices read on each vendor's pricing page on October 26, 2026.

Price point	Devin	Claude	Cursor
Free	$0, light agent quota, unlimited Tab completions	$0, chat and apps, no Claude Code	$0, limited Agent and Chat requests
Entry paid tier	Pro: frontier models from four providers, Devin Cloud, up to 10 concurrent sessions	Pro, the same monthly price, or $17 a month on the annual plan, Claude Code included	Plus, $8 a month, Grok models only; Premium, $30 a month, third-party models at API list price on top
Top individual tier	Max: weekly quota only, no daily cap	Ultra (formerly Max 20x): 10 times Pro's weekly usage, no five-hour limit for verified accounts	Ultra: 50 times the Plus pool, third-party models still billed separately
Overage	Extra usage at API pricing, credits never expire	Optional usage credits at standard API rates	Prepaid credit for third-party models at API list price
Team seat	One full seat per developer plus an account minimum	Team seats at the Pro monthly price per seat, billed annually	Premium, Super or Ultra seat at the individual price

The three ladders reset on different clocks. Devin resets daily and weekly on Pro, weekly on Max. Claude's pricing page says every plan resets on a rolling five-hour session window, with weekly limits on paid plans, and that Claude Code draws from the same pool as the chat apps. Cursor's pricing page sells a usage pool that scales from Plus, and bills Claude, GPT and Gemini separately on every plan. If you want Claude's models inside your agent at a flat price, Devin Pro is the only one of the three that includes them without a per-token line.

The trade-off runs the other way on model choice. Claude Ultra is Claude only; Devin Max is any frontier model in the picker, with Cognition's SWE-2 free on top. Cursor Ultra is the most usage for Grok and Composer (see Grok pricing for what the Grok side of that bundle costs on its own) and the least predictable for anything else. For the Cursor and Copilot ladders in full, read Cursor vs GitHub Copilot; for the Claude side, Claude pricing and is Claude Max worth it. The AI subscription prices page puts all three next to 37 AI tools.

Toolradar data: visitors to Devin's Toolradar page compared it most often with Manus and Claude Code over the 30 days to late October 2026, which says the buyer's real question is agent versus agent, not agent versus editor. Manus is a general-purpose agent with credit-based plans (see Manus AI pricing); if what you want is an app builder rather than an agent on your own repository, Lovable vs Replit covers that pair.

Four quirks that change the bill

• Your ChatGPT plan can pay for GPT models. Cognition's ChatGPT plan documentation says a linked ChatGPT Go, Plus or Pro subscription pays for GPT model requests in your sessions, on Pro, Max and Teams, and that usage does not count against the Devin quota. Non-GPT models, fast and priority GPT variants, cloud machine time and code scans still draw on Devin. See ChatGPT pricing for what those plans cost.
• Windows sessions cost about 9% more. The usage documentation says Windows sessions consume roughly 9% more usage than Linux sessions; macOS sessions cost the same as Linux under promotional launch pricing that may change.
• Idle sessions stop metering. Devin sleeps after 30 minutes of inactivity (5 minutes for Lite and API-created sessions) and consumes nothing while asleep beyond the units that keep the VM alive. Waiting on you, on a test suite or on a repository clone does not count either.
• Public pull requests are reviewed free. Anyone can run Devin Review on a public GitHub PR at devinreview.com without an account or credits; private repositories need a plan.

Methodology

Prices and plan contents were read on devin.ai/pricing and windsurf.com/pricing on October 26, 2026, and the billing mechanics from Cognition's self-serve, enterprise, usage and ChatGPT-plan documentation on the same day. Past prices come from dated Internet Archive copies of Devin's own pricing page, linked where used. Claude and Cursor figures come from their pricing pages, read the same day. Totals marked as our arithmetic multiply printed prices; we never estimate a price the vendor does not print. The Toolradar figures come from the software directory we run, with their snapshot dates. Prices are re-checked when the page is refreshed and the updated date moves only on a real change.

FAQ

How much does Devin cost in 2026?

Devin costs $0 on Free, $20 a month on Pro and $200 a month on Max for a single user, and $80 a month plus $40 per full developer seat on Teams for up to 200 users. Enterprise is priced on a quote in Agent Compute Units. Every paid plan includes a usage quota, and extra usage past it is billed at API pricing.

Is Devin free?

Yes, with limits. The Free plan is $0 and includes a light quota to code with agents, limited model availability, unlimited inline edits and unlimited Tab completions in the desktop editor. It does not include Devin Cloud, DeepWiki, the Devin API or the Slack, Linear and GitHub integrations, which start on Pro.

Does Devin still use ACUs?

Only on Enterprise. The self-serve Free, Pro, Max and Teams plans moved to daily and weekly usage quotas with on-demand credits for overage between late March and late April 2026, replacing the Core plan at $2.25 per ACU and the Team plan at $500 a month. Enterprise contracts are still billed in ACUs at the rate in the order form.

What is the difference between Devin Pro and Max?

Both are single-user plans with the same features and up to 10 concurrent sessions. Pro has a daily and a weekly quota; Max, at ten times the price, has a significantly larger weekly quota with no daily cap. Max suits bursty, parallel agent work; Pro suits steady daily use, with extra usage purchasable at API pricing when needed.

Is Windsurf pricing the same as Devin pricing?

Yes. Windsurf's editor is now Devin Desktop, and windsurf.com/pricing shows the same Free, Pro, Max, Teams and Enterprise plans at the same prices as devin.ai/pricing. One subscription covers the desktop editor, the Devin CLI and Devin Cloud sessions from a single quota.

How does Devin Teams pricing work for a small team?

Teams has a monthly minimum equal to two full seats. A full seat carries its own Pro-equivalent quota and Devin Desktop; a flex seat is free and draws on shared on-demand credits. Two full seats cost exactly the minimum, five cost two and a half times it, and with fewer than two full seats the rest of the minimum is charged as prepaid credits the whole team can use.

Cite this: AIpresso, "Devin Pricing 2026: Free, Pro, Max and Teams Explained", November 2026.

AIpresso covers the AI tools worth your time in a 5-minute morning read. Join professionals who want the signal without the hype.

Some offers on this page may be paid placements or contain affiliate links.