NGDBench - NGD-Prime Management sMAPE: leaderboard
Metric: Symmetric mean absolute percentage error (0 to 2) of the numeric answers to the validation queries run after each step of NGDBench's five-step management sessions on the NGD-Prime graph (the model writes each create, update or delete edit in Cypher), against the answers on the correctly edited graph; lower is better. Source: arxiv.org. Saturation forecast: Around January 2027. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Coder 480B A35B Instruct | 0.31 | #302 |
| 2 | DeepSeek V3.2 | 0.35 | #198 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=ngdbench-ngd-prime-management-smape · How It Works · Data refreshed daily, snapshot 2026-10-11.