IG-Bench - IG-Exam: leaderboard

Metric: Exact accuracy (%; every required field correct) on the closed-form lineage-reasoning questions of IG-Exam (1,029 instances in 42 task types over 1,961 expert-audited scientific lineage traces in 10 domains). Source: arxiv.org. Saturation forecast: Around 2030. 12 models tracked.

Top models

#ModelScore
1GPT-5.523.1
2Qwen 3.6 Max Preview20.6
3Kimi K2 (Thinking)20.4
4Gemini 3.1 Pro (Preview)20.1
5Claude Opus 4.719.3
6DeepSeek V4 Pro17.9
7GLM-5.117.7
8MiniMax-M2.711.9

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