UniQL - Spark: leaderboard
Execution accuracy on the Spark dialect of UniQL (2026), which ports the 1,534 BIRD development questions to 16 SQL dialects: the percent of questions whose generated Spark query returns the gold result. Higher is better.
Metric: Execution accuracy (%) on the Spark 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 October 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 56.65 | #266 |
| 2 | Claude Sonnet 4.5 | 56.13 | #250 |
| 3 | GPT-5.1 Codex | 53.65 | #288 |
| 4 | GPT-5 Mini | 52.93 | #348 |
| 5 | DeepSeek V4 Flash | 51.56 | #258 |
| 6 | Qwen 3 32B | 49.22 | #960 |
| 7 | Qwen 3 8B | 47.46 | #1830 |
| 8 | Qwen 3 4B | 43.94 | #2281 |
| 9 | Llama 3 70B Instruct | 42.89 | #1676 |
| 10 | GPT-3.5 Turbo | 39.37 | #2753 |
| 11 | Qwen 3 1.7B | 35.27 | #5520 |
| 12 | Llama 3 8B Instruct | 23.34 | #4612 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=uniql-spark · How It Works · Data refreshed daily, snapshot 2026-11-13.