IG-Bench - IG-Exam - Genome Abstraction: leaderboard
Metric: Exact accuracy (%) on the 125 genome-abstraction instances (5 task types) of IG-Exam. Source: arxiv.org. Saturation forecast: Around 2030. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 32.4 |
| 2 | Claude Opus 4.7 | 28.5 |
| 3 | GLM-5.1 | 28.5 |
| 4 | Kimi K2 (Thinking) | 27.9 |
| 5 | GPT-5.5 | 27.5 |
| 6 | Qwen 3.6 Max Preview | 26.9 |
| 7 | DeepSeek V4 Pro | 23.9 |
| 8 | MiniMax-M2.7 | 22.9 |
Interactive version: theaggregate.ai/benchmark?slug=ig-bench-ig-exam-genome-abstraction · How It Works · Data refreshed daily, snapshot 2026-09-29.