Read about our latest announcements.
Devin now automatically breaks large tasks into stacks of small, independently reviewable PRs — and handles the rebasing, conflict resolution, and CI verification across the stack for you.

Kimi K3 is now live in Devin Desktop and Devin CLI. Optimized for long horizon agentic coding tasks, it approaches frontier-level performance on FrontierCode 1.1 — surpassing GPT-5.5 and behind only Opus, Fable, and GPT-5.6 Sol — with particular strength on debugging.

Claude Opus 5 is now live in Devin Desktop and Devin CLI, and will be part of Devin Cloud’s mode mix. On FrontierCode 1.1, it approaches Fable-level performance at half the cost, with particular strength on difficult debugging and root-cause analysis tasks.
Give Devin the ability to run on any machine.

GPT-5.6 models are now live in Devin. These models stand out for combining top-tier coding-agent performance with very strong token/cost efficiency.

Claude Fable 5 is now available in Devin Cloud’s Ultra agent, Devin Desktop, and Devin CLI.

A new architecture from Cognition for scaling full-codebase reasoning with distributed agents

How we built an eval of real, recent vulnerabilities to measure Security Swarm — and why it finds more bugs at lower cost than any tool we tested.

Claude Sonnet 5 is now live in Devin Desktop and Devin CLI. It pairs frontier-level coding performance with a more affordable price point, using roughly 30% less quota than Claude Sonnet 4.6 through August 31, 2026.

Kimi K2.7 and GLM 5.2 are now available in Devin Desktop and CLI. Both models deliver strong FrontierCode Extended scores and won't consume quota for Pro, Max, and Teams plans until July 5.

Every pull request in Devin Review now gets a security review. Devin catches auth bypasses and logic flaws that pattern-matching scanners miss, with full remediation from finding to fix.

The next generation of Windsurf: a full IDE with the Agent Command Center built in for managing fleets of local and cloud agents from one surface.