K12-Bench - Cross-Chapter Indexing: leaderboard

The Cross-Chapter Indexing task of K12-Bench (2026), built from the K12-KGraph knowledge graph of Chinese K-12 mathematics, physics, chemistry and biology textbooks, as four-option multi-select questions answered zero-shot: locate where knowledge first appears and which chapters a chapter depends on. Exact match in percent (every option must be chosen correctly); higher is better.

Metric: Cross-Chapter Indexing exact match (%, all-or-nothing over the selected option set) on four-option multi-select questions generated from the K12-KGraph curriculum knowledge graph of Chinese K-12 mathematics, physics, chemistry and biology textbooks; zero-shot, answer-only, temperature 0: locate where knowledge first appears and the prerequisite chapters of a chapter (8,624 items); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash81.7#175
2Gemma 4 31B (IT)73.4#350
3Gemini 2.5 Flash72.8#463
4Qwen 3 32B72.1#972
5GPT-5.270.1#185
6Ministral-3-14B-Instruct59.3#1139
7GPT-4o55.6#751
8GPT-5 Mini55.5#364
9GLM-4.7 Flash48.9#1349
10Llama 3 8B Instruct11.5#4852

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=k12-bench-cross-chapter-indexing · How It Works · Data refreshed daily, snapshot 2026-11-21.