BesiegeField - Catapult: leaderboard
Metric: Best-scaffold Mean Score. Source: besiegefield.github.io. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 9.83 |
| 2 | O3 | 9.14 |
| 3 | Llama 4 Scout | 3.59 |
| 4 | DeepSeek V3 | 3.5 |
| 5 | Claude Opus 4 (20250514) | 2.27 |
Interactive version: theaggregate.ai/benchmark?slug=besiegefield-catapult · How It Works · Data refreshed daily, snapshot 2026-09-05.