IG-Bench - IG-Exam - Evolutionary Reasoning: leaderboard
IdeaGene-Bench (IG-Bench, 2026) represents papers as typed Idea Genome objects aligned across 1,961 scientific lineage traces in 10 domains. This board covers the 409 IG-Exam instances that infer the dynamics, driver and object fates of a transition, scored by exact accuracy in percent; higher is better.
Metric: Exact accuracy (%) on the 409 evolutionary-reasoning instances (17 task types) of IG-Exam. Source: arxiv.org. Saturation forecast: Around 2029. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 23.3 | #49 |
| 2 | Kimi K2 (Thinking) | 19.6 | #439 (Kimi K2) |
| 3 | Qwen 3.6 Max Preview | 18.8 | #159 |
| 4 | DeepSeek V4 Pro | 18.6 | #188 |
| 5 | Gemini 3.1 Pro (Preview) | 17.8 | #94 |
| 6 | GLM-5.1 | 17.4 | #179 |
| 7 | Claude Opus 4.7 | 17.1 | #79 |
| 8 | MiniMax-M2.7 | 10.9 | #518 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ig-bench-ig-exam-evolutionary-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-13.