We cry about AI tools so you don't have to.

Pricing Watch

Windsurf 2.0 Bundled Devin. Here's What That Actually Costs.

Windsurf 2.0 added Devin to Pro plans on April 15, but the shared quota promise doesn't mean free.

Updated
pricingWindsurfDevinIDEAI coding
Disclosure: Tool Crier may earn a commission if you buy through links in this article. We paid for these tools ourselves or tested via free trials. Commissions don't influence what we write.

Windsurf’s launch post on April 15 buried the lede: Devin is “included in Pro” — but that single word, “included,” glosses over two completely different billing models depending on whether you’re a self-serve user or on an enterprise contract. We read the quota documentation and cross-referenced it with real ACU cost data so you can decide whether to actually use this feature or let it sit.

What shipped: Devin inside the editor

Windsurf 2.0 hands off local Cascade sessions to a cloud Devin agent with one click. Devin then handles debugging, testing, deployment, and longer tasks on its own VM — meaning your laptop can close and the work keeps running. It’s a genuinely useful hand-off: you sketch the plan locally, then turn it over to a 24/7 cloud agent to finish.

The catch is the billing. TestingCatalog’s coverage confirms that self-serve and enterprise users hit completely different pricing paths.

Self-serve: Shared quota, but not what “shared” means

If you’re on Windsurf Pro ($20/month), Max, or Teams, Devin consumes your existing Windsurf quota. That sounds free — and for minimal Devin use it mostly is. New GitHub connections get up to $50 in extra credits, so your first few Cloud Devin sessions land on Cognition’s dime.

But here’s where the announcement gets sticky: when you exceed your Windsurf quota (50 premium interactions per day on Pro), you’re billed at live API pricing for both Windsurf use and the Devin work that consumes that quota. The bill doesn’t itemize it separately — it just pulls from the same bucket.

Enterprise: ACUs on top of your Windsurf bill

Enterprise customers don’t use the shared quota model. Devin requires “admin enablement and separate purchase of Cognition Platform access.” Translation: if you’re on a $200+/month enterprise Windsurf contract, Devin is not automatically included — you’re negotiating a separate Cognition deal.

Those deals use Agent Compute Units (ACUs). One ACU ≈ 15 minutes of active Devin work. The Core plan charges $2.25/ACU; Teams charges $2.00/ACU after 250 included units. A medium-complexity task (say, adding auth to an API or writing a full test suite) runs 10–25 ACUs — so $22–56 on Core pricing alone.

Three usage profiles: what you’ll actually pay

Light user: ~5–10 Devin sessions/month, mostly bug fixes or small features (1–3 ACUs each).

  • Self-serve Pro: Fits in your standard quota most months; overage is $0–20 depending on other use.
  • Enterprise: $25–75/month in ACUs, on top of your Windsurf contract.

Active solo dev: 15–20 Devin sessions/month, mix of debugging and medium features (5–10 ACUs average).

  • Self-serve Pro: Likely hits quota ceiling by week 3–4. Overage runs $40–80/month depending on concurrent Windsurf usage.
  • Enterprise: $150–300/month in ACUs, plus Windsurf fees.

Small team: 40–60 Devin sessions/month across 3–4 developers, heavier workloads.

  • Self-serve Teams: Quota exhausts quickly; you’re looking at $150–300/month overage or switching to Max (unknown tier pricing, but assume 2–3x Pro quota for ~$50–60/month).
  • Enterprise: $400–800/month in ACUs, negotiated separately from Windsurf.

The pattern: the more you use Devin, the faster you hit quota limits and the more your bill accelerates.

Where the announcement misled you

“Included in Pro” is technically true but operationally false. Devin is bundled into the plan, not the pricing — it consumes quota you’ve already paid for, but it also competes with every other Windsurf AI use for those same tokens. If you’re an active Windsurf user already, Devin doesn’t add “free” work; it adds expensive overage. If you’re light on Windsurf use, Devin stays cheap until suddenly it doesn’t.

Enterprise users get the worst surprise: the announcement implies Devin is included in their plans, but it’s a separate line item requiring separate procurement from Cognition. The DEV Community thread flags this exact issue — users discovered “the usage has been severely reduced” and “pricing has been made more opaque” compared to older credit models.

Who should use it, who should skip it

Use Devin in Windsurf if: you’re a light to moderate Windsurf user (hitting quota only 1–2x/month) and you have specific, discrete tasks you want automated (full test suites, deployment scripting, refactoring a single module). The hand-off is smooth, and you’ll stay under quota limits.

Avoid it if: you’re already a power Windsurf user. Devin will push you past quotas fast, and the billing jumps from “$20/month” to “$100+/month” without warning. For enterprise teams, negotiate the Cognition pricing separately before enabling — don’t assume it’s baked into your Windsurf contract.

What we’re watching next

Windsurf pricing flipped from credits to quotas in March and the Devin bundling is the first major feature stress-test of that new model. It’s also part of the broader May 2026 IDE pricing churn hitting the whole coding-assistant space right now. If shared-quota billing becomes a pattern for feature rollouts, self-serve users will hit ceilings faster, and the “$20/month” entry point will look increasingly misleading. We’re tracking whether Cognition publishes pre-task ACU estimates (they currently don’t) and whether enterprises start bundling Devin into flat-rate contracts instead of ACU-based billing. Either change would reshape the actual cost.

For now: read your quota docs, track your Devin usage for 30 days before committing to heavy use, and don’t trust “included” without checking your bill.

← More Pricing Watchs

What we don't know is documented at the end of this article. We update when we learn more.