Benchmarks and Comparisons?

[JohnSmithyy]

Can you provide any benchmarks or comparisons to other tools and code review tools?

Here is an example: https://github.com/Ataraxy-Labs/inspect?tab=readme-ov-file#code-review-benchmark

[tirth8205]

Internal benchmarks already exist and can be reproduced with code-review-graph eval --all. Results are in evaluate/reports/summary.md. Key numbers: 5-24x token reduction vs full-file context, 100% impact recall, sub-millisecond search. Head-to-head comparisons against other tools are not available yet.

[tirth8205]

We now have an eval framework — run code-review-graph eval --all to benchmark against real repos. Reports are generated under evaluate/reports/. See the README for details on methodology.

[tirth8205]

Issue correctly closed as a general discussion question. The eval command (code-review-graph eval) provides benchmarking capabilities for measuring graph quality.