UniQL - Druid: leaderboard

Metric: Execution accuracy (%) on the Druid 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 2030. 13 models tracked.

Top models

#ModelScore
1Claude Sonnet 4.539.9
2GPT-5.1 Codex38.2
3GPT-5 Mini37.74
4Gemini 2.5 Pro37.42
5DeepSeek V4 Flash34.22
6Qwen 3 32B33.83
7Qwen 3 8B31.29
8Llama 3 70B Instruct26.92
9Qwen 3 4B26.53
10GPT-3.5 Turbo22.75
11Qwen 3 1.7B18.71
12Llama 3 8B Instruct16.56

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