All companies
technology

Windsurf pricing

devin.ai facts checked analysis reviewed
Quick summary
In this page
AI Summary
  • Windsurf is an agentic AI coding IDE, formerly Codeium, now owned by Cognition and rebranded Devin Desktop, with windsurf.com/pricing 308-redirecting to devin.ai/pricing.
  • Windsurf pricing runs a free $0 tier, Pro at $20 per month, Max at $200 per month, a Teams plan at $80 per month plus $40 per full dev seat, and a custom-quoted Enterprise tier.
  • Each paid plan includes a usage quota that refreshes on a daily and weekly basis, and agent message cost varies by the model used, task size, and reasoning required.
  • Usage beyond the included quota is purchased at underlying model API pricing rather than a flat per-message fee.
  • In 2026 Windsurf replaced its long-running Cascade prompt-credit system with daily and weekly quotas and raised Pro from $15 to $20, drawing public user backlash.
  • Windsurf's ownership split in July 2025 when OpenAI's $3 billion acquisition collapsed, Google paid $2.4 billion to hire the founders, and Cognition bought the remaining company.
Pricing summary
Windsurf 2026 — seat + daily/weekly usage quota (now Devin Desktop)
Hybrid: per-seat subscription bundles a usage allowance that refreshes daily and weekly; overage billed at raw model API pricing.
Free
$0
Individuals trying the agent
10× Pro
Max
$200 /mo
Heaviest individual agent users
Teams
$80 /mo
Collaborating dev teams
Enterprise
Let's talk
Large orgs needing SSO + VPC
Windsurf is now Devin Desktop (Cognition). Prices read from devin.ai/pricing on 2026-06-08. Each paid plan includes a usage quota that refreshes daily/weekly; overage is billed at API pricing.

About

Windsurf is an agentic AI coding IDE, originally launched as Codeium by Exafunction in 2022 and best known for a free-forever autocomplete tool that grew into the agentic Windsurf Editor (powered by the Cascade agent) in November 2024. The company rebranded from Codeium to Windsurf in April 2025. It competes head-to-head with Cursor, GitHub Copilot, Replit and Tabnine in the AI-coding-tool category, and by mid-2025 it reported roughly $82M in ARR and 350+ enterprise customers.

Windsurf is also the subject of one of the most dramatic corporate sagas in AI. In July 2025, OpenAI’s reported $3 billion acquisition of Windsurf collapsed when its exclusivity window expired (Microsoft’s blanket IP rights over OpenAI’s acquisitions were a reported sticking point). The same day, Google paid roughly $2.4 billion in a reverse-acquihire to hire founders Varun Mohan and Douglas Chen plus key researchers into Google DeepMind, leaving the company behind. Days later, on 2025-07-14, Cognition — the maker of the autonomous engineering agent Devin — acquired the remaining Windsurf company, including its product, brand, IP and ~250-person team.

Under Cognition the product has been folded into Devin Desktop: windsurf.com/pricing now 308-redirects to devin.ai/pricing, where the desktop IDE is positioned as the local front-end to Devin’s cloud agents (Devin Cloud). In-editor Tab completions and inline edits pair with cloud agent sessions metered against a usage quota. For the most current pricing, see the Devin pricing page.


Pricing summary : How Windsurf’s pricing model works

Windsurf (now Devin Desktop, under Cognition) runs a freemium subscription with a usage quota: a free $0 tier, two individual paid steps (Pro $20/mo and Max $200/mo), a Teams plan at $80/mo for the team plus $40/mo per full developer seat, and a custom-quoted Enterprise tier. Every paid plan ships with a usage allowance that refreshes automatically on a daily and weekly basis; cost per agent message varies by the model used, task size, and reasoning required, and any usage beyond the included quota is purchased at API pricing.

The metered dimensions are:

  • Seats — Pro and Max are per-individual; Teams bills $80/mo base plus $40/mo per full dev seat (each full user carries its own quota).
  • Usage quota (agent messages / flow actions) — included credits refresh daily and weekly; spend depends on model, task complexity, and reasoning.
  • Overage at API pricing — once the quota is exhausted, extra usage is metered at underlying model API rates (a pass-through token cost), not a flat per-message fee.

What makes this different: the quota refreshes on a rolling daily/weekly basis rather than a single monthly bucket, and overage is billed transparently at API pricing instead of marked-up credits — a hybrid pricing model blending seat-based pricing with a credit-style usage allowance. This is the fourth distinct metering model Windsurf has shipped in roughly 16 months.


Pricing by product

Devin Desktop (Individual plans)

TierPriceIncludedKey mechanics
Free$0Light quota to code with agents; limited model availability; unlimited inline edits and Tab completionsSelf-serve download; entry funnel
Pro$20 / moEverything in Free, plus increased quotas, full model availability (OpenAI, Claude, Gemini frontier + free SWE 1.6), Devin Cloud agents”Popular” tier for individual devs; overage at API pricing
Max$200 / moEverything in Pro, plus significantly higher quotasFor power users who exhaust Pro quota

Devin Desktop (Team plans)

TierPriceIncludedKey mechanics
Teams$80 / mo + $40 /seat/moEverything in Pro, plus unlimited team members, shared collaboration, centralized billing, admin dashboard with analytics, priority support$80/mo team base plus $40/mo per full dev seat; each full user carries its own quota
EnterpriseLet’s talk (custom)Everything in Teams, plus highest-priority support, dedicated account management, SAML/OIDC SSO, centralized enterprise admin controls, dedicated/VPC deploymentSales-led; contact via Cognition sales form

Sales motions across products: PLG / self-serve for Free, Pro, Max, and Teams (sign up and pay online); sales-led for Enterprise (contact-sales quote).


Hidden costs : What Windsurf users actually pay

The headline Pro price ($20/mo) understates what heavy agent users actually pay, because the included quota is a soft cap: once exhausted, every additional agent message is billed at the underlying model’s raw API price. Cost per message also varies by model, task size and reasoning, so a single $20 line can hide a wide overage tail.

Archetype 1 — a solo developer who leans on Cascade daily. A Pro user ($20/mo) who runs the agent through most of the workday will typically stay inside the daily/weekly quota for routine work, but a few large refactors or long agent sessions push past it and into API-priced overage.

Line itemMonthly cost
Pro seat$20
Overage agent usage (≈2–3 heavy days past quota, at model API rates)$25–$60
Estimated total$45–$80

So the “$20 IDE” can land closer to a $50–$80 bill in heavy weeks — and unlike the old credit system, unused daily/weekly allowance no longer banks for those bursts.

Archetype 2 — a 10-person team on Teams. Teams is an $80/mo team base plus $40/mo per full dev seat, and each full seat carries its own daily/weekly quota with overage at API pricing.

Line itemMonthly cost
Team base$80
10 full dev seats × $40$400
Team-wide overage (a few seats exceeding quota at API rates)$100–$300
Estimated total$580–$780

The per-seat $40 plus the $80 base means the platform fee alone is $480/mo before any agent usage — and because each seat meters independently, a handful of power users can drive the overage line above the entire base. This is the classic bill-shock and cost-unpredictability failure mode that the quota model was meant to tame but partly re-introduces.

Want to estimate your own Windsurf bill? Use the Windsurf pricing calculator to model your monthly cost based on seats, quota consumption, and API-priced overage.


Pricing evolution : Windsurf pricing history and changes

Cadence

QuarterPrice changesProduct / SKU additionsNotes
2023 Q201Codeium adds an Enterprise (VPC/on-prem) tier; Individual stays free forever
2024 Q311Paid Teams tier at $15/seat/mo introduced; still flat seats, no usage meter
2024 Q412Windsurf Editor launches (2024-11-13) with the Cascade agent; Cascade dual-credit system (User Prompt + Flow Action) goes live 2024-12-06 — Free $0, Pro $15, Pro Ultimate $60
2025 Q211Codeium rebrands to Windsurf (Apr 2025); Pricing v2 on 2025-04-21 eliminates Flow Action credits (prompt-credits only); Teams $35 / Teams Ultimate $90 added
2025 Q300Ownership saga: OpenAI deal collapses, Google acquihire, Cognition acquires Windsurf (2025-07-14); pricing unchanged short-term
2026 Q1–Q211Credits replaced by daily/weekly quotas (between 2026-03-08 and 2026-04-01); Pro $15→$20, new Max $200 tier; later folded into Devin Desktop (windsurf.com → devin.ai)

Tracked range: 2023-01–2026-06 (Wayback codeium.com + windsurf.com snapshots). Quarters not listed were verified stable.

Notable changes

  • 2023-04 — Codeium adds an Enterprise tier (Contact for Pricing, VPC/on-prem, local fine-tuning); Individual remains free forever.
  • 2024-09 — First paid self-serve tier: Teams $15/seat/mo (admin dashboard, GPT-4/Claude 3.5 Sonnet). Still a flat subscription.
  • 2024-11-13Windsurf Editor launches as an agentic IDE with the Cascade agent.
  • 2024-12-06Cascade dual-credit system: Free $0 (5+5 credits), Pro $15 (500 prompt + 1,500 flow-action), Pro Ultimate $60.
  • 2025-04-21Pricing v2 (windsurf.com/blog/pricing-v2): Flow Action credits eliminated; prompt-credits only; add-ons $10/250 with rollover; a 127-point Hacker News thread followed.
  • 2025-07-11 — OpenAI’s reported $3B acquisition collapses; Google pays ~$2.4B to hire founders Mohan & Chen into DeepMind (1,055-point Hacker News thread on The Verge’s report).
  • 2025-07-14Cognition acquires Windsurf (~$82M ARR, 350+ enterprise customers, ~250-person team).
  • 2026 Q1–Q2 — Prompt credits replaced by daily/weekly usage quotas; Pro $15→$20; new Max $200 tier; product folded into Devin Desktop.

The credit-to-quota change and the ownership saga in detail

Two trust events define Windsurf’s pricing history.

The ownership split (July 2025). Windsurf was reportedly hours from a $3 billion sale to OpenAI when the exclusivity period lapsed on 2025-07-11. Within hours Google announced a ~$2.4 billion deal to license Windsurf’s technology and hire founders Varun Mohan and Douglas Chen into DeepMind — a “reverse-acquihire” that took the leadership but not the company. Head of business Jeff Wang became interim CEO. Three days later Cognition acquired the rest. The Verge’s report drew a 1,055-point, 685-comment Hacker News discussion (2025-07-11), one of the largest AI-company threads of the year. The pricing surface itself didn’t move during the drama, but ownership of the meter changed hands twice in a week.

The credit-to-quota change (2026). Windsurf’s most contentious pricing event was the removal of its Cascade prompt-credit system in favor of daily/weekly quotas. Screenshot evidence shows windsurf.com/pricing still running the credit model on 2026-03-08 (Pro $15, 500 credits/mo) and the new quota model live by 2026-04-01 (Pro $20, allowance “refreshes daily & weekly”, overage “at API price”, a new Max $200 tier). Critically, the new quotas do not roll over: a bursty developer who codes hard a few days a week and rests the rest now watches unused daily/weekly allowance evaporate instead of banking purchased credits. Reddit users called it a “33% price hike” and documented effective-usage drops of roughly 4x; Cognition’s leadership framed quotas as necessary to support long-running agents and prevent extreme usage spikes from instantly draining accounts.


What’s unique : Windsurf’s distinctive pricing mechanics

1. Overage at raw API pricing, not marked-up credits. Once a paid plan’s quota is exhausted, additional usage is consumed “at API pricing” — the underlying model provider’s wholesale token rate — rather than at a marked-up credit price. That is unusually transparent for the category: most competitors meter overage in proprietary credits whose dollar value is deliberately fuzzy. It also means Windsurf’s overage margin is thin by design, betting that the seat fee, not the meter, carries the economics.

2. A quota that refreshes daily AND weekly, not monthly. Where almost every credit-based billing competitor resets a single monthly bucket, Windsurf’s allowance refreshes on a rolling daily and weekly cadence. For steady everyday users this smooths spend; for bursty developers it is the source of the 2026 backlash, because unused allowance cannot be banked for an intense sprint. The cadence is the differentiator and the liability at once.

3. The Cascade agent and the “flow action” meter it retired. Cascade — Windsurf’s planning-executing-reviewing agent — was the first IDE agent to meter “flow actions” (the agent’s own tool calls) separately from user prompts. When reasoning models started averaging ~4 tool calls per prompt, that dual meter became un-spendable, and Windsurf killed Flow Action credits in April 2025. The rise and fall of flow-action billing is a live case study in metering an agent’s autonomous steps.

4. The pricing meter changed owners mid-stream. Few pricing pages in this corpus have been operated by four different entities — Exafunction (Codeium), Windsurf Inc., (briefly) a Google-licensed shell, and Cognition — inside 18 months. The Devin Desktop consolidation means Windsurf’s seat-plus-quota model now has to coexist with Devin’s separate ACU (Agent Compute Unit) cloud-agent meter, a rare two-meter household under one brand.


Strengths & weaknesses

StrengthsWeaknesses
Overage billed at transparent raw API pricing, not opaque marked-up creditsDaily/weekly quotas don’t roll over — punishes bursty, part-time developers
Free $0 tier and $20 Pro keep the entry funnel competitive with Cursor and CopilotPricing model has changed four times in ~16 months — low predictability for buyers
Daily/weekly refresh smooths spend for steady, everyday agent users2026 credit-to-quota switch raised Pro 33% ($15→$20) and cut effective usage (~4x reports)
Teams plan is self-serve with centralized billing and admin analyticsTeams $80 base + $40/seat means a high platform floor before any usage
Backed by Cognition (Devin), with frontier model access (OpenAI, Claude, Gemini)Brand churn (Codeium→Windsurf→Devin Desktop) and redirect create buyer confusion
Agent (Cascade) and overage economics are unusually legible vs. credit-pool peersPer-message cost varies by model/task/reasoning — hard to forecast a monthly bill

Billing UX : Windsurf billing controls and transparency

  • Self-serve plan selection — Free downloads directly; Pro, Max, and Teams are selected and paid online via “Select plan” on devin.ai/pricing.
  • Daily/weekly refreshing usage quota — each paid plan’s included allowance resets automatically on a daily and weekly basis rather than a single monthly bucket (per the pricing-page FAQ).
  • Extra usage at API pricing — when included usage is exhausted on a paid plan, additional usage is purchasable and consumed at underlying model API rates.
  • Teams admin dashboard with analytics — centralized billing plus an admin dashboard surfacing team usage analytics on the Teams plan.
  • Enterprise centralized admin controls — SAML/OIDC SSO, RBAC, and centralized admin controls for managing multiple Devin organizations, plus VPC/hybrid deployment for billing and data isolation.
  • Referral credits — historically, referring a friend to a paid plan earned bonus credits (e.g. 250–500 flex/add-on credits), a self-serve growth lever wired into the billing system.

Strategic wins : Why Windsurf’s pricing decisions worked

1. Free-forever individual tier built an enormous top-of-funnel

For years Codeium gave individuals unlimited autocomplete for free while competitors charged, turning a generous free tier into the cheapest possible distribution channel. That base of hundreds of thousands of daily active users — not the meter — is what made the company worth a reported $3B to OpenAI and ~$2.4B to Google. The lesson is classic usage-based pricing: a credible free tier can be the most valuable line item on the page even though it generates $0 directly.

2. Overage at raw API pricing built trust the credit-pool model erodes

By pricing overage at the model provider’s wholesale API rate rather than a marked-up internal credit, Windsurf made the most-scrutinized number on its pricing page legible. In a category where buyers are increasingly burned by opaque credit conversions, transparent pass-through pricing is a differentiator — see how AI companies are moving from per-user licenses to usage. It signals confidence that the seat fee, not meter margin, carries the model.

3. Aggressively simplifying the meter when it stopped working

When reasoning models pushed flow-action consumption past what users could spend, Windsurf killed the dual-credit system outright in April 2025 rather than patching it. Choosing the right usage metric and being willing to retire one that no longer maps to value is a discipline most teams lack. It also pre-empted a worse version of the predictability complaints that later hit the quota model.

4. Capturing frontier-model access as a paid-tier gate

The Pro tier’s headline benefit is access to OpenAI, Claude and Gemini frontier models plus Windsurf’s own SWE models — a clean, legible reason to upgrade from Free. Gating model quality (not just quota size) gives the $20 step a value story beyond “more of the same,” which is exactly what the shift toward outcome- and capability-based pricing rewards.


Areas to improve : Gaps in Windsurf’s pricing approach

1. Let quota roll over — or offer a banking option

The single biggest complaint about the 2026 quota model is that unused daily/weekly allowance evaporates, punishing bursty and part-time developers and producing the “33% price hike” framing. A concrete fix: let a portion of unused weekly allowance bank for a few weeks, or offer a “rollover” toggle the way the old credit system did. This directly addresses the cost-unpredictability and bill-shock reaction that drove the backlash.

2. Publish a credits-to-quota conversion so buyers can compare

Because the meter changed shape (credits → daily/weekly allowance described only as “Light / Standard / Heavy”), buyers can’t tell whether $20 buys more or less than the old $15/500-credit plan. A published, concrete mapping — e.g. “Standard ≈ N prompts/day across these models” — would defuse the “they cut 95% of usage” claims. Vague capacity labels invite the worst-case interpretation; specifics from a usage-metering and tracking standpoint would rebuild trust.

3. Stabilize and communicate the brand/redirect

Four owners and three names (Codeium → Windsurf → Devin Desktop) in 18 months, plus a silent 308-redirect from windsurf.com to devin.ai, leave buyers unsure what they’re even purchasing. A clear, dated migration notice on both domains — what changes, what your existing plan maps to, what stays — would reduce churn risk during the Devin consolidation.

4. Add usage caps and budget alerts for API-priced overage

Since overage is billed at raw API rates with “unlimited” extra usage, a single runaway agent session can generate an unbounded bill. Hard spend caps and proactive budget alerts (see thresholding and alerting) would convert the transparency advantage into genuine cost safety for teams.


Key takeaways

  1. Changing your value metric is the riskiest pricing move you can make. Windsurf changed its core meter four times in ~16 months, and the change that altered the shape of the bucket (rolling-credits → non-rolling daily/weekly quota) — not the price increase itself — generated the backlash. When you change what you count, customers re-evaluate the whole relationship.
  2. A 33% price increase is survivable; removing rollover is not. The $15→$20 bump alone would have been routine; pairing it with loss of bankable allowance is what produced “4x effective usage cut” claims. Predictability is often valued more than absolute price — protect the mechanic customers plan around.
  3. Transparent overage pricing buys trust you can spend later. Billing overage at raw API rates made Windsurf’s most-scrutinized number defensible and gave it credibility headroom — credibility it partly drew down with the opaque “Light/Standard/Heavy” quota labels.
  4. A free-forever tier can be your most valuable asset even at $0 revenue. Codeium’s free base is what made the company a multi-billion-dollar acquisition target three times over. Distribution and DAU count can dominate enterprise valuation in this category.
  5. Vague capacity labels invite worst-case interpretation. Replacing concrete “500 credits” with “Standard” let users assume the largest possible cut. If you must abstract the meter, publish a concrete conversion so buyers can’t fill the gap with the worst number.

UBP implications

  1. Agentic products break single-meter assumptions. Windsurf’s flow-action meter died because reasoning models made the agent’s own tool calls outnumber user prompts ~4:1. As agents take more autonomous steps per request, metering the steps becomes un-spendable; metering the request (or the outcome) is more durable. Expect more vendors to retire step-level meters.
  2. Refresh cadence is itself a pricing lever — and a risk. Moving from a monthly bucket to daily/weekly refresh changes who wins and loses without changing the headline price. UBP designers should treat reset cadence as a first-class decision: it determines whether bursty or steady usage profiles are subsidized, and switching it mid-stream re-prices every customer silently.
  3. Pass-through (raw-cost) overage is emerging as a trust signal. Windsurf’s “at API pricing” overage shows a path where vendors compete on transparency by charging wholesale for marginal usage and taking margin on the seat. For outcome- and capability-based pricing, separating “platform margin” from “pass-through compute” is becoming a credibility marker buyers actively look for.

Sources


Bottom line

Windsurf’s pricing is a moving target wrapped in a corporate thriller: a free-forever Codeium autocomplete grew into the Cascade-powered Windsurf Editor, survived a three-way OpenAI/Google/Cognition split in a single week, and now sits inside Devin Desktop charging $20 Pro against a daily/weekly quota with raw-API overage. The mechanics are unusually transparent for the category — but four metering changes in 16 months, and a 2026 switch that removed credit rollover, make predictability the model’s weakest point.

Want to compare Windsurf against Cursor, GitHub Copilot, and other AI coding tools? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Windsurf is now Devin Desktop — windsurf.com redirects to devin.ai

windsurf.com/pricing 308-redirects to devin.ai/pricing under the Devin brand. Live plans: Free $0, Pro $20/mo, Max $200/mo, Teams $80/mo team base + $40/mo per full dev seat, Enterprise custom. Each paid plan includes a daily/weekly-refreshing quota; overage billed at API pricing.

Windsurf is now Devin Desktop — windsurf.com redirects to devin.ai - windsurf.com/pricing 308-redirects to devin.ai/pricing under the Devin brand. Li
captured

Credits killed for daily/weekly quotas; Pro $15→$20, Max $200 added

Between 2026-03-08 (last credit-model snapshot) and 2026-04-01, Windsurf replaced prompt credits with a usage allowance that 'refreshes daily & weekly', with extra usage billed 'at API price'. Pro rose to $20/mo, a new Max tier at $200/mo appeared, and Teams moved to $40/user/mo. The shift drew public backlash over lost predictability for bursty coders.

Credits killed for daily/weekly quotas; Pro $15→$20, Max $200 added - Between 2026-03-08 (last credit-model snapshot) and 2026-04-01, Windsurf replace
captured

Prompt-credit model under Cognition

Windsurf.com/pricing (now © Cognition) ran prompt credits: Free $0 (25 credits/mo), Pro $15/mo (500), Teams $30/user/mo (500/user), Enterprise (1,000/user, custom). SSO add-on +$10/user/mo. A 'joining forces with Cognition' banner sat atop the page.

Prompt-credit model under Cognition - Windsurf.com/pricing (now © Cognition) ran prompt credits: Free $0 (25 credits/m
captured

Cognition acquires Windsurf (after OpenAI/Google saga)

Days after OpenAI's reported $3B acquisition collapsed (2025-07-11) and Google paid ~$2.4B to hire founders Mohan and Chen into DeepMind, Cognition (maker of Devin) acquired the remaining Windsurf — IP, product, brand, ~$82M ARR, 350+ enterprise customers and a ~250-person team. Pricing structure unchanged short-term.

Pricing v2 — flow-action credits eliminated

Windsurf dropped Flow Action credits entirely; users now pay only for User Prompts regardless of how many steps Cascade takes. Pro stayed $15/mo for 500 prompt credits; add-ons $10/250 with monthly rollover and auto-refills; early adopters grandfathered at $10/mo for a year. Announced on windsurf.com/blog/pricing-v2; a 127-point Hacker News thread followed.

Rebrand: Codeium becomes Windsurf

The company and IDE consolidated under the Windsurf brand; pricing moved to windsurf.com/pricing. Team tiers expanded to Teams $35 and Teams Ultimate $90/user/mo, with an Enterprise SaaS tier; the Cascade dual-credit model carried over.

Rebrand: Codeium becomes Windsurf - The company and IDE consolidated under the Windsurf brand; pricing moved to wind
captured

Cascade dual-credit system (User Prompt + Flow Action)

Pricing moved from flat seats to credits. Free $0 (5 User Prompt + 5 Flow Action credits), Pro $15/mo (500 prompt + 1,500 flow-action credits), Pro Ultimate $60/mo (infinite prompt + 3,000 flow-action). Add-on credits $10/300. First true usage meter.

Cascade dual-credit system (User Prompt + Flow Action) - Pricing moved from flat seats to credits. Free $0 (5 User Prompt + 5 Flow Action
captured

Windsurf Editor launches with Cascade agent

Codeium launched the Windsurf Editor, an agentic IDE built on VS Code with the Cascade agent. Cascade introduced metered 'flow' work, setting up the credit model that landed weeks later.

Teams tier at $15/seat introduced

Individual remained $0/forever with unlimited autocomplete and chat. A paid Teams plan appeared at $15/seat/mo (admin dashboard, seat management, GPT-4/Claude 3.5 Sonnet, Forge code review). Enterprise stayed custom. Still flat subscription — no usage credits.

Teams tier at $15/seat introduced - Individual remained $0/forever with unlimited autocomplete and chat. A paid Team
captured

Enterprise tier added (VPC / on-prem)

Individual stayed Free Forever; a new Enterprise tier (Contact for Pricing) added VPC/on-prem deployment, local fine-tuning, usage stats and priority support. Two-tier free-PLG-into-enterprise model.

Enterprise tier added (VPC / on-prem) - Individual stayed Free Forever; a new Enterprise tier (Contact for Pricing) adde
captured

Codeium: one free-forever plan

codeium.com/pricing showed a single plan — 'Early users get free forever access to Codeium': unlimited code suggestions, 20+ languages, VS Code and JetBrains extensions, Discord support. No paid tier.

Codeium: one free-forever plan - codeium.com/pricing showed a single plan — 'Early users get free forever access
captured
Trivia
  • · Windsurf's company split three ways in July 2025: OpenAI's reported $3B acquisition collapsed, Google paid ~$2.4B to hire the founders into DeepMind, and Cognition bought the remaining product, brand and ~250-person team — all within days.
  • · Codeium's pricing started as literally 'one plan' in 2023 — free forever — and stayed free for individuals until the Cascade credit system arrived in December 2024.
  • · Windsurf has changed its core metering model three times in ~16 months: flat seats → dual User-Prompt + Flow-Action credits (Dec 2024) → prompt-credits-only (Apr 2025) → daily/weekly quotas (2026).

Questions & answers

What is Windsurf's pricing model?
Windsurf (now Devin Desktop, under Cognition) uses a freemium subscription with a usage quota: a free $0 tier plus paid plans that each include a daily/weekly-refreshing usage allowance, with overage billed at model API pricing.
Does Windsurf offer a free tier?
Yes. The Free plan is $0 and includes a light quota to code with agents, limited model availability, and unlimited inline edits and Tab completions.
How much does Windsurf cost per month?
On devin.ai/pricing, Pro is $20/mo, Max is $200/mo, and Teams is $80/mo plus $40/mo per full dev seat. Enterprise is custom-quoted via Cognition sales.
Why did Windsurf change its pricing in 2026?
In early 2026 Windsurf replaced its Cascade prompt-credit system with daily and weekly usage quotas and raised Pro from $15 to $20/mo. Cognition framed quotas as necessary to support long-running agents, but the change drew user backlash over lost predictability.
What happened in the OpenAI, Google, and Cognition deal?
On 2025-07-11 OpenAI's reported $3B acquisition of Windsurf collapsed; the same day Google paid ~$2.4B to hire founders Varun Mohan and Douglas Chen into DeepMind. Days later, on 2025-07-14, Cognition (maker of Devin) acquired the remaining Windsurf company, product, and team.
How does Windsurf compare to Cursor and GitHub Copilot on price?
Windsurf Pro ($20/mo) sits alongside Cursor Pro ($20/mo) and GitHub Copilot Pro ($10/mo). Windsurf differentiates with a daily/weekly quota that refreshes rather than a single monthly bucket, and overage billed at raw model API pricing.