EdgeBench - Code: leaderboard

Metric: Systems and software engineering tasks: mean best-so-far score (0-100) after a 12-hour continuous agent run with evaluator feedback, three runs per task, averaged over the family's tasks; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 5 models tracked.

Top models

#ModelScore
1Claude Opus 4.867.4
2GPT-5.565
3GPT-5.454.1
4GLM-5.150.9
5DeepSeek V4 Pro43

Interactive version: theaggregate.ai/benchmark?slug=edgebench-code · How It Works · Data refreshed daily, snapshot 2026-09-29.