IG-Bench - IG-Arena: leaderboard
IdeaGene-Bench (IG-Bench, 2026) represents papers as typed Idea Genome objects aligned across 1,961 scientific lineage traces in 10 domains. IG-Arena has each system write research proposals for 30 frontier questions and scores them by the Population-Evolution Score, the mean of heredity, variation and selection ratings from a three-judge panel on a 0-100 scale; higher is better.
Metric: Population-Evolution Score (0-100), the mean of heredity, variation and selection scores from a three-judge panel for proposals inserted into a given lineage population, 30 frontier tasks in 10 domains, pooled over the question-only, library and lineage information settings). Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 86.5 | #49 |
| 2 | DeepSeek V4 Pro | 82.7 | #188 |
| 3 | Claude Opus 4.7 | 82.6 | #79 |
| 4 | Gemini 3.1 Pro (Preview) | 82.1 | #94 |
| 5 | Kimi K2 (Thinking) | 81.1 | #439 (Kimi K2) |
| 6 | GLM-5.1 | 80.7 | #179 |
| 7 | Qwen 3.6 Max Preview | 79.8 | #159 |
| 8 | MiniMax-M2.7 | 70.5 | #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-arena · How It Works · Data refreshed daily, snapshot 2026-11-13.