Devin 3 hits 90% on SWE-bench Verified — Cognition completes Windsurf acquisition at $250M and bundles Devin inside the IDE

Cognition's Devin 3 model now clears 90% on SWE-bench Verified — the first SWE-bench score consistently above the 90% threshold from any autonomous engineering agent. Cognition has completed its acquisition of Windsurf (the remaining stake after Google's earlier $2.4B acqui-hire of the founders) for $250M. The combination bundles Devin Cloud and Devin Terminal CLI inside the Windsurf IDE; Windsurf Pro raised to $20/month with a new $200/month Max tier.

The SWE-bench-90 milestone is the autonomous-engineering tier's headline number. Cursor Composer 2.5's Build in Parallel is the strongest IDE-bundled autonomous-engineering offer at the per-token-pricing end of the market; Windsurf+Devin is now the strongest quota-bundled autonomous-engineering offer at the subscription end. The market has settled into a two-vendor split with materially different pricing models.

For procurement teams, the choice is increasingly operational rather than capability-bound. Teams that prefer fixed-quota predictable budgeting go Windsurf+Devin. Teams that prefer per-token elasticity with model-routing flexibility go Cursor. Cursor's $0.50/$2.50 pricing and MS Teams integration from the AM cycle remains the strongest counter-positioning. Both are now genuinely defensible.

Lushbinary — AI coding agents comparison 2026 → · Kingy AI — Codex vs Claude Code vs Cursor vs Windsurf vs Manus → · Shareuhack — Cursor vs Claude Code vs Windsurf →