PredicateLongBench - Clustered Near-Sorted Decoys (Synthetic): leaderboard

Metric: Accuracy (%; share of the 100 synthetic word lists of about 128K tokens whose answer is exactly the one word sequence satisfying the predicate, locate the sorted run with 8 near-sorted decoys clustered next to it; up to 16K output tokens including reasoning, an answer past the cap counts as wrong). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.

Top models

#ModelScore
1Claude Opus 4.6 (Thinking)98
2GPT-5.4 (High)70
3Gemini 3.1 Pro (Preview) (High)64
4Claude Opus 4.630
5MiniMax-M2.710
6GPT-5.4 (Non-reasoning)5
7Gemini 3.1 Pro (Preview) (Low)5
8GLM-5.12
9Qwen 3.5 397B A17B0

Interactive version: theaggregate.ai/benchmark?slug=predicatelongbench-clustered-near-sorted-decoys-synthetic · How It Works · Data refreshed daily, snapshot 2026-09-29.