KnotBench - Same Canonical PD (Images): leaderboard

The A3-I task of KnotBench (2026, The Gordian Knot for VLMs): judge whether two diagrams have the same canonical PD code, with answers checked against Regina canonical knot signatures. The score is accuracy in percent under strict answer matching, with a random baseline of 50 percent; higher is better.

Metric: Accuracy (%) on the 200 A3-I items: whether two diagrams have the same canonical PD code; zero-shot with a forced ANSWER line and strict matching, 64K output-token budget, random baseline 50%; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 2 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 4.790.5#85
2GPT-5Thinking74.5#178

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

Interactive version: theaggregate.ai/benchmark?slug=knotbench-same-canonical-pd-images · How It Works · Data refreshed daily, snapshot 2026-11-21.