Natural Language to Mongosh: leaderboard
MongoDB text-to-query benchmark evaluating natural-language generation of mongosh queries with execution, output, normalization, latency, and token metrics.
Metric: NeXMaNeR (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.7 Sonnet | 90.32 | #249 |
| 2 | Claude 3.5 Haiku | 87.65 | #568 |
| 3 | Gemini 2.0 Flash | 87.52 | #342 |
| 4 | O3 Mini | 86.29 | #272 |
| 5 | GPT-4o | 86.2 | #343 |
| 6 | Gemini 2.5 Pro (Preview 03-25) | 85.66 | #123 |
| 7 | GPT-4o Mini | 82.9 | #599 |
| 8 | Nova Pro (v1) | 82.61 | #497 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=natural-language-to-mongosh · How It Works · Data refreshed daily, snapshot 2026-10-20.