Graphwalks parents <128k: leaderboard
Graphwalks parent-retrieval task below 128k context, measuring whether models can recover predecessor relationships from serialized graphs.
Metric: Score (%). Source: llm-stats.com. Status: saturation imminent. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 | 89.8 |
| 2 | GPT-5.2 | 89 |
| 3 | GPT-5 | 73.3 |
| 4 | GPT-4.5 | 72.6 |
| 5 | GPT-5.4 Mini | 71.5 |
| 6 | GPT-4.1 Mini | 60.5 |
| 7 | O3 Mini | 58.3 |
| 8 | GPT-4.1 | 58 |
| 9 | GPT-5.4 Nano | 50.8 |
| 10 | GPT-4o | 35.4 |
| 11 | GPT-4.1 Nano | 9.4 |
Interactive version: theaggregate.ai/benchmark?slug=graphwalks-parents-under-128k · How It Works · Data refreshed daily, snapshot 2026-09-05.