UniQL - ClickHouse: leaderboard

Metric: Execution accuracy (%) on the ClickHouse 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 August 2028. 13 models tracked.

Top models

#ModelScore
1Claude Sonnet 4.556.84
2Gemini 2.5 Pro53.98
3GPT-5.1 Codex52.61
4GPT-5 Mini51.56
5Qwen 3 32B49.74
6DeepSeek V4 Flash48.04
7Qwen 3 8B46.54
8Qwen 3 4B43.02
9Llama 3 70B Instruct40.16
10GPT-3.5 Turbo35.14
11Qwen 3 1.7B33.9
12Llama 3 8B Instruct20.8

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