BIRD-SQL — leaderboard
Database-grounded text-to-SQL benchmark over large cross-domain databases, scored primarily by whether generated SQL executes to the correct answer.
Metric: Execution Accuracy (%). Source: bird-bench.github.io. Status: saturation imminent. 55 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (xHigh) | 72.55 |
| 2 | Claude Opus 4.6 | 70.15 |
| 3 | Claude Sonnet 4.5 | 66.85 |
| 4 | Command A | 65.68 |
| 5 | GLM-4.7 | 62.94 |
| 6 | DeepSeek R1 | 60.93 |
| 7 | Kimi K2 (Thinking) | 59.87 |
| 8 | GPT-4 | 54.89 |
| 9 | Claude 2 | 49.02 |
Interactive version: theaggregate.ai/benchmark?slug=bird-sql · How the rankings work · Data refreshed daily, snapshot 2026-07-22.