Windsurf pricing: plans, quotas, and what changed

Windsurf is billed on Cognition's Devin plan sheet: Free at $0, Pro at $20 per month, Max at $200 per month, Teams at $80 per month for the team plan plus $40 per month per full dev seat, and Enterprise at custom pricing, in USD on monthly billing 1. Prompt credits are gone: paid plans carry a usage allowance that refreshes daily and weekly, extra usage beyond it is sold at API pricing on paid plans, and the numeric limits the vendor publishes are estimate ranges, not ceilings 2. Two figures frame the choice: Max costs 10 times Pro for about six times the estimated allowance, and Teams stacks an $80 base fee on top of its $40 seat fee.

What each Windsurf tier costs, and what each one adds

The base tier does the flat work everywhere below it: Free costs $0 and includes a light quota to code with agents, limited model availability, and unlimited inline edits and Tab completions 1. Every paid tier is that base plus a delta:

• Pro, $20 per month: adds increased quotas with access to frontier models from OpenAI, Claude, Gemini, and SpaceXAI plus leading open source models, full model availability, free use of the SWE-2 model, access to cloud agents in Devin Cloud, and the option to buy extra usage at API pricing 1
• Max, $200 per month: adds one thing the page names, "significantly higher quotas" 1
• Teams, $80 per month for the team plan plus $40 per month per full dev seat: adds unlimited team members, centralized billing, an admin dashboard with analytics, and priority support 1
• Enterprise, custom pricing: adds SAML/OIDC SSO, centralized enterprise admin controls, dedicated account management, and dedicated deployment options 1

One dated promo rides the shared sheet: Pro and above include Devin Desktop and the CLI free through October 10, 2026 1.

Prompt credits are gone: what the March 2026 restructure changed

• Windsurf previously billed plans in prompt credits, charging the same credits for a quick question and a long complex task; Cognition says many users felt pressured to cram requests into single prompts, ultimately degrading the quality of their Windsurf experience 2
• The replacement was announced March 18, 2026 and took effect March 19, 2026: "industry-standard quotas," an allowance refreshing daily and weekly, with extra usage on paid plans consumed at API pricing 2
• Existing Pro and Teams subscribers were grandfathered at their current price indefinitely while moving to allowance billing, and each paid subscriber received a free extra week to trial the new system 2
• Previously purchased add-on credits were converted into a dollar amount of extra usage at the rate originally paid 2
• Windsurf's own pricing URL now serves the Devin plan sheet: Devin Cloud, Ask Devin, and SWE-2 all bill from the same page 1

The quota numbers are estimates, not published limits

The plan page publishes no numeric quota at all; its FAQ describes only an allowance that refreshes on a daily and weekly basis 1. The only numbers the vendor has put out are the announcement's estimated messages per day, and they swing by model class 2:

• Premium Plus models (Opus 4.6, GPT-5.4, GPT-5.3-Codex): 7 to 27 messages per day on Pro and Teams, 42 to 170 on Max 2
• Premium models (e.g. Sonnet 4.6, GPT-5.2, Gemini Pro): 8 to 101 per day on Pro and Teams, 47 to 631 on Max 2
• Lightweight models (e.g. Haiku, Flash): 47 to 190 per day on Pro and Teams, 291 to 1,190 on Max 2

Only the Premium Plus list is closed; the Premium and Lightweight lists are illustrative, not exhaustive 2. An additional weekly limit applies on top of the daily window, and neither the plan page nor the announcement publishes its value 1 2. The vendor calls the daily figures estimates based on current usage patterns 2. A write-up quoting exact prompt counts as Windsurf limits is reading these ranges as ceilings, or predates the March 2026 switch.

Max costs 10 times Pro for about six times the estimated allowance

Max is the pure quota tier: the plan page lists nothing on top of Pro except higher quotas 1. The arithmetic the vendor does not run: $200 against $20 is 10 times the price, while the announcement's own estimate bands put Max at 5.9 to 6.3 times Pro's daily allowance on every model class, at both ends of every range (42 over 7 is 6.0; 1,190 over 190 is 6.3) 2. Per unit of estimated headroom, Max runs about 1.6 times Pro's rate. Treat it as ceiling insurance, priced against the vendor's own statement that most users never hit their limits 1, and remember the ratios compare estimate bands, not published quota units.

The cheaper capacity lever sits inside Pro. On the vendor's estimates, lightweight models deliver roughly seven times the daily message count of Premium Plus ones (47 to 190 against 7 to 27) 2. If the task tolerates a lighter model, switching class moves your allowance further than the $180 jump to Max does, at no cost.

Teams carries an $80 base the announcement table did not show

The live plan page bills Teams at $80 per month for the team plan plus $40 per month per full dev seat 1. The March announcement's plan table lists Teams at $40 per seat per month and shows no base fee 2. The two vendor surfaces disagree on the base; budget from the live page, which is the current price of record.

At the live numbers 1:

• 5 developers: $280 per month
• 10 developers: $480 per month
• 25 developers: $1,080 per month
• 50 developers: $2,080 per month

Per developer that works out to $56, $48, $43.20, and $41.60, a curve that approaches but never reaches $40, double Pro's per-seat price, at any team size; the arithmetic is the $40 seat fee plus the $80 base divided by headcount. Individual Pro seats cost $20 each flat, so the same headcounts run $100, $200, $500, and $1,000 per month 1. Teams buys the management layer (centralized billing, the analytics dashboard, priority support), not cheaper seats 1.

Extra usage at API pricing, in practice

When the allowance runs out, paid plans buy extra usage that draws down a dollar balance at API price for the models used; cost per message varies with the model, the task size, and the reasoning required, and priority or speed variants consume the allowance faster 2 1. Free models like SWE-1.5 keep working beyond the limit at no charge 2. With no published ceiling there is no published worst-case bill: the first month on any tier is a measurement month. Teams admins can watch the analytics dashboard; individuals have no usage meter on either vendor surface, only the announcement's estimate bands as a guide 1 2. The vendor's own stretch advice is to trim prompt context and drop to smaller models such as Haiku, GPT 5.2 Mini, or Kimi K2.5 1.

Cursor and Copilot price the same job with published meters

• GitHub Copilot's paid plans start at $10 per user per month with $15 of monthly AI credits (a Free tier sits at $0 below Pro on the same page); Pro+ is $39 with $70 in credits and at least 4 times Pro's included usage; Max is $100 with $200 in credits and at least 2.9 times Pro+ usage, and additional credits can be purchased on paid plans 3. Windsurf's Max tier, at $200, is double Copilot Max 1
• Cursor's Pro plan is $20 per month, the same as Windsurf Pro, and Cursor Teams starts at $40 per user per month on the Standard tab, with no base fee; the page also lists a Premium variant whose price it does not print. Beyond the included usage, on-demand use is billed in arrears 4
• The structural read: all three ladders share one shape, an included allowance plus metered overage, but the transparency differs. Copilot prints the meter in dollars, Cursor points to its docs, and Windsurf publishes no quota numbers at all, only estimate ranges with API-priced overflow 3 4 1

Which Windsurf plan is for whom

Free fits trying agent-driven editing while keeping unlimited edits and completions. Pro is the default for daily frontier-model work; if you exhaust allowances, try the lighter-model lever before the $180 upgrade. Max is for sustained Premium Plus sessions where roughly six times the headroom beats paying API rates for overflow. Teams is for organizations that need the billing and admin layer and will pay at least double Pro's per-seat price for it. Enterprise is for SSO, compliance, and dedicated deployment 1. One group should move carefully: subscribers grandfathered at their pre-restructure price keep it indefinitely per the announcement 2.

This page is maintained against the vendor's plan page on a refresh cycle, and pricing changes are logged in the ProvenBrief newsletter. Verified September 2026 against https://windsurf.com/pricing and Cognition's March 18, 2026 announcement.