CAME-Bench - Large: leaderboard

Metric: Answer-set macro F1 (%; 61 questions on two trajectories averaging 408K tokens; full context, left-truncated to the model's window, gpt-4.1-mini judge). Source: arxiv.org. Saturation forecast: Around January 2028. 5 models tracked.

Top models

#ModelScore
1GPT-4.1 Mini21.3
2GPT-5 Mini (Medium)21.2
3GPT-4o Mini1.7
4Qwen 3 235B A22B 2507 Instruct1.6
5DeepSeek V3.10

Interactive version: theaggregate.ai/benchmark?slug=came-bench-large · How It Works · Data refreshed daily, snapshot 2026-09-26.