DBLifeBench - Progressive Text2SQL: leaderboard
Progressive-Text2SQL, the operation-phase task introduced by DBLifeBench: a complex query is decomposed into a reasoning graph of described sub-queries that the model writes step by step, scored by the mean node execution accuracy in percent; higher is better.
Metric: Graph execution accuracy (%; mean node execution accuracy over 1,149 Progressive-Text2SQL reasoning graphs, each node a sub-query with its own description; temperature 0.3, top-p 0.2, SQLite). Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o Mini | 54.31 | #1623 |
| 2 | GPT-4o | 53.79 | #741 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=dblifebench-progressive-text2sql · How It Works · Data refreshed daily, snapshot 2026-11-13.