ProLLM - SQL Disambiguation: leaderboard
ProLLM benchmark measuring LLM ability to disambiguate and generate correct SQL from ambiguous natural language queries.
Metric: Score (%). Source: www.prollm.ai. Saturation forecast: Not forecast. 52 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.5 | 53.1 | #274 |
| 2 | O1 | 49.2 | #183 |
| 3 | Grok 2 | 47.9 | #452 |
| 4 | GPT-3.5 Turbo | 47.7 | #865 |
| 5 | GPT-4.1 | 46.4 | #248 |
| 6 | Qwen 2.5 72B Instruct | 46 | #454 |
| 7 | DeepSeek V3 | 44.7 | #321 |
| 8 | Gemini 2.0 Flash | 42 | #342 |
| 9 | Mistral Small 3.1 | 40.8 | #610 |
| 10 | Command A | 40.6 | #508 |
| 11 | Grok 3 | 40.3 | #304 |
| 12 | GPT-4.1 Mini | 40.1 | #355 |
| 13 | GPT-4 Turbo | 37.5 | #404 |
| 14 | Nova Pro | 37.4 | #569 |
| 15 | GPT-4o | 36.7 | #343 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=prollm-sql-disambiguation · How It Works · Data refreshed daily, snapshot 2026-10-20.