PlanarBench: leaderboard
PlanarBench (2026) asks a model to draw an ASCII layout of a planar graph, given only its edge list, with every node placed once and no edge crossings, over 199 non-isomorphic connected planar graphs with 2 to 7 vertices. Each graph scores 1 for a strictly valid straight-line drawing or half credit each for valid node placement and valid edge connectivity, summed out of 199; higher is better.
Metric: Score (out of 199): sum over the 199 planar graphs (2-7 vertices) of the per-graph drawing score, 1 for a strictly valid straight-line crossing-free ASCII drawing, otherwise half credit each for a coordinate-valid node placement and for BFS-valid edge connectivity; one attempt per graph, single fixed prompt; the paper prints its top 21 of 91 configurations; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 151 | #113 |
| 2 | Claude Opus 4.5 (Thinking) | 144 | #156 (Claude Opus 4.5) |
| 3 | GPT-5 Codex | 130 | #245 |
| 4 | GPT-5 Codex (High) | 127 | #245 (GPT-5 Codex) |
| 5 | GPT-5 | 124.5 | #167 |
| 6 | GPT-5 (High) | 124 | #167 (GPT-5) |
| 7 | Claude Sonnet 4.5 (Thinking) | 112.5 | #250 (Claude Sonnet 4.5) |
| 8 | GPT-5 Mini (High) | 111 | #348 (GPT-5 Mini) |
| 9 | GPT-5.2 (Low) | 101.5 | #176 (GPT-5.2) |
| 10 | Gemini 2.5 Pro (Preview 05-06) | 96 | #275 |
| 11 | Claude 3.7 Sonnet (Thinking) | 95.5 | #458 (Claude 3.7 Sonnet) |
| 12 | O4 Mini | 93.5 | #353 |
| 13 | O1 | 91 | #393 |
| 14 | Gemini 2.5 Pro (Preview 03-25) | 85 | #236 |
| 15 | GLM-4.7 | 84 | #366 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=planarbench · How It Works · Data refreshed daily, snapshot 2026-11-13.