Coding · sub-raceNarrow lead
Code Review & Security race
Spotting real bugs and vulnerabilities in diffs while keeping false positives low.
- Leader
- Opus 5.5
- Margin
- 0.7 pts
- 0.8% ahead
- Weeks at #1
- 4
- in a row
- Changes
- 6
- in 2026
Warming up the track…
Oct 19, 2026Opus 5.5 leads
2025Apr ’25Jul ’25Oct ’252026Apr ’26Jul ’26Oct ’26
Score history
Top 6 today over the leader's frontier · click to jump to a week
#1 frontierOpus 5.5Fable 5.1GPT-6 AstraGPT-6.1 SolGemini 4 ArgonSonnet 5.5
Reigns at #1
Longest: GPT-5, 15 weeks (from Aug 2025).
Sibling races
More coding races
Why this race matters
A reviewer that cries wolf gets ignored. Precision matters as much as recall.
How it is scored
F1 of flagged issues against seeded and historical bugs.
Unit: percentHigher is betterPhoto finish under 0.5 ptsDominant over 3.0 pts
Data provenance
CyberGym
Vulnerability reproduction tasks on real projects.
Review precision evals
Seeded-bug diffs scored on precision and recall.
Scores here are illustrative placeholders that mimic real benchmark patterns. Swap in live data through the score provider. Methodology