Ironbee Raw Code Review - Full - F2: leaderboard
Ironbee raw code review (2026): standalone model review of real pull requests, three repeated runs. Full uses 50 PRs and Frontier 10 PRs; no cohort pooling. Model-specific requested tiers remain explicit; precision, recall, F1 and F2 are separate metrics. Aggregated source fractions are multiplied by 100, with full precision retained. F2 (%).
Metric: F2 (%). Source: raw.githubusercontent.com. Saturation forecast: Rough model projection: around 2026. 52 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Ironbee Study claude-sonnet-5.5 - low tier | 64.38 | |
| 2 | Ironbee Study claude-opus-5.5 - low tier | 62.14 | |
| 3 | Ironbee Study claude-opus-5 - low tier | 61.92 | |
| 4 | Ironbee Study grok-4.7 - low tier | 61.87 | |
| 5 | Ironbee Study claude-opus-5.5 - default tier | 61.86 | |
| 6 | Ironbee Study claude-sonnet-5.5 - default tier | 61.75 | |
| 7 | Ironbee Study claude-sonnet-5.5 - xhigh tier | 61.13 | |
| 8 | Ironbee Study muse-spark-1.3 - default tier | 60.88 | |
| 9 | Ironbee Study grok-4.7 - default tier | 60.67 | |
| 10 | Ironbee Study claude-opus-5.5 - xhigh tier | 60.36 |
Interactive version: theaggregate.ai/benchmark?slug=ironbee-raw-code-review-full-f2 · How It Works · Data refreshed daily, snapshot 2026-10-21.