# 2026 AI Code Review Benchmark

> Precision, recall and F1 of eight AI code reviewers (Entelligence, Codex, Claude Code, Bugbot, Greptile, CodeRabbit, Copilot, Graphite) against a golden set of 67 review comments on real pull requests from Cal.com, Sentry, Discourse, Keycloak and Grafana.

Source: https://entelligence.ai/code-review-benchmark-2026
Updated: 2026-05-15

Published 2026-01-15, last reviewed 2026-05-15. Quote these figures with that date and this page as the source.

## Results

| Rank | Tool | F1 | Recall | Precision | Golden comments found |
| --- | --- | --- | --- | --- | --- |
| 1 | Entelligence | 47.2% | 44.8% | 50.0% | 30/67 |
| 2 | Codex | 45.4% | 40.3% | 51.9% | 27/67 |
| 3 | Claude | 42.8% | 43.3% | 42.3% | 29/67 |
| 4 | Bugbot | 39.4% | 46.3% | 34.4% | 31/67 |
| 5 | Greptile | 36.9% | 46.3% | 30.7% | 31/67 |
| 6 | CodeRabbit | 33.0% | 49.2% | 24.8% | 32/65 |
| 7 | Copilot | 22.6% | 34.3% | 16.8% | 23/67 |
| 8 | Graphite | 13.4% | 7.5% | 66.7% | 5/67 |

Download: [results.csv](https://entelligence.ai/code-review-benchmark-2026/results.csv)

## Entelligence by repository

| Repository | F1 | Recall | Precision | Golden comments found |
| --- | --- | --- | --- | --- |
| Cal.com | 68.1% | 81.2% | 58.6% | 13/16 |
| Grafana | 50.0% | 33.3% | 100.0% | 3/9 |
| Sentry | 45.5% | 29.4% | 100.0% | 5/17 |
| Keycloak | 38.5% | 33.3% | 45.5% | 4/12 |
| Discourse | 30.3% | 38.5% | 25.0% | 5/13 |

## Methodology

Golden comments come from an ensemble of frontier LLMs (Claude Opus, Gemini 2.5, GPT-5) given each diff plus its LSP-traced dependencies, reduced by majority vote with GPT-4o as arbiter, then validated by human reviewers. Each tool is scored on how many golden comments it finds (recall) and how many of its comments match one (precision).

Evaluated pull requests and tool comments: https://github.com/AI-Code-Review-Evals