SelfBench - PostHog/posthog: leaderboard
Metric: Task success (%) on 36 tasks rebuilt from the repository's merged pull requests (hidden tests). Source: selfbench.dev. Saturation forecast: Around October 2027. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 Astra (High) | 69.44 |
| 2 | Claude Opus 5.5 (High) | 69.44 |
| 3 | GPT-6.1 Sol (High) | 63.89 |
| 4 | Claude Opus 5.5 (Low) | 61.11 |
| 5 | GPT-6 Sol (High) | 55.56 |
| 6 | Kimi K3 (High) | 47.22 |
| 7 | GPT-6 Luna (High) | 38.89 |
| 8 | GPT-6 Sol (Low) | 38.89 |
| 9 | GPT-6 Luna (Low) | 27.78 |
Interactive version: theaggregate.ai/benchmark?slug=selfbench-posthog-posthog · How It Works · Data refreshed daily, snapshot 2026-10-03.