Augment Agent consistently outperforms the competition in code quality and reliability. Don’t take our word for it: we’re now #1 on the SWE-Bench leaderboard at 65.4%—and we’re aiming even higher.

Augment's retrieval system dominates in real-world code understanding, not just artificial benchmarks. Achieving 81.8% keyword recall on our repository-aware benchmark—outperforming top models from Salesforce and Amazon by nearly 2x.

We achieve 67% accuracy on the CCEval test—much better than GitHub Copilot's 50% (with all files open) and 30% (without open files). This means better, smarter code suggestions for you.