UniQL - Drill: leaderboard

Metric: Execution accuracy (%) on the Drill dialect of UniQL (1,534 BIRD development-set questions with human-verified SQL in 16 aligned dialects): given the target dialect, schema and question, the model writes one query whose execution result must match the gold result; inference-only, one run; higher is better. Source: arxiv.org. Saturation forecast: Around September 2028. 13 models tracked.

Top models

#ModelScore
1Claude Sonnet 4.552.28
2Gemini 2.5 Pro50.85
3GPT-5 Mini48.83
4GPT-5.1 Codex48.11
5Qwen 3 32B46.54
6DeepSeek V4 Flash46.28
7Qwen 3 8B44.13
8Qwen 3 4B41.72
9Llama 3 70B Instruct37.74
10GPT-3.5 Turbo29.2
11Qwen 3 1.7B27.57
12Llama 3 8B Instruct18.45

Interactive version: theaggregate.ai/benchmark?slug=uniql-drill · How It Works · Data refreshed daily, snapshot 2026-09-29.