PredicateLongBench - Lexicographic Longest (LongBench): leaderboard

Metric: Accuracy (%; share of the 93 LongBench v2 documents under 170K tokens used as word lists whose answer is exactly the one word sequence satisfying the predicate, retrieve the longest lexicographically sorted run; up to 16K output tokens including reasoning, an answer past the cap counts as wrong). Source: arxiv.org. Saturation forecast: Around 2029. 9 models tracked.

Top models

#ModelScore
1Claude Opus 4.6 (Thinking)20
2Gemini 3.1 Pro (Preview) (Low)10
3GPT-5.4 (High)7
4Claude Opus 4.65
5Qwen 3.5 397B A17B0
6GLM-5.10
7MiniMax-M2.70
8Gemini 3.1 Pro (Preview) (High)0
9GPT-5.4 (Non-reasoning)0

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