Agent Security League - Functional Correctness: leaderboard
Agent Security League measures functional correctness for coding-agent/model pairs. Scores are retained from each framework run, with date and raw agent identity; independent per-metric maxima are not synthesized across agents. Existing downstream harness selection remains separate from extraction.
Metric: Functional Correctness (%). Source: www.endorlabs.com. Saturation forecast: Around January 2027. 32 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 | 87.2 | #9 |
| 2 | GPT-5.5 | 84.9 | #31 |
| 3 | Claude Sonnet 5 | 83.2 | #57 |
| 4 | GLM-5.2 | 82.5 | #73 |
| 5 | GPT-6 Astra | 82.1 | #6 |
| 6 | Gemini 3.5 Flash | 79.3 | #60 |
| 7 | Claude Opus 4.6 | 78.2 | #63 |
| 8 | GPT-6.1 Sol | 77.7 | #7 |
| 9 | Claude Opus 4.7 | 76.5 | #49 |
| 10 | Composer 2.5 | 75.4 | #122 |
| 11 | Claude Opus 4.8 | 73.7 | #37 |
| 12 | Claude Opus 5 | 73.7 | #12 |
| 13 | Gemini 3.1 Pro (Preview) | 73.2 | #58 |
| 14 | GPT-6 Sol | 72.1 | #14 |
| 15 | GPT-5.6 Sol | 70.9 | #19 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=agent-security-league-functional-correctness · How It Works · Data refreshed daily, snapshot 2026-10-20.