BesiegeField - Catapult - Iterative Editing: leaderboard
Metric: Mean distance (m). Source: besiegefield.github.io. Saturation forecast: Around December 2026. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 | 9.14 |
| 2 | Qwen 3 Coder 480B A35B | 5.1 |
| 3 | Gemini 2.5 Pro | 4.67 |
| 4 | DeepSeek V3 | 3.07 |
| 5 | Llama 4 Scout | 1.28 |
| 6 | Claude Opus 4 (20250514) | 1.18 |
Interactive version: theaggregate.ai/benchmark?slug=besiegefield-catapult-iterative-editing · How It Works · Data refreshed daily, snapshot 2026-10-09.