Tinybird AI SQL Benchmark - Exactness: leaderboard
Tinybird's LLM SQL benchmark has each model write SQL for 50 natural-language analytical questions over a 200M-row GitHub Archive table in Tinybird. Exactness is a 0-100 correctness score from comparing each query's output with human-written reference queries; higher is better. Runs the site still shows with an unvalidated exactness of 0 are omitted.
Metric: Result exactness vs human reference queries (0-100). Source: llm-benchmark.tinybird.live. Saturation forecast: Around 2029. 398 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 65.48 | #51 |
| 2 | Claude Fable 5 | 65.05 | #20 |
| 3 | Muse Spark 1.3 | 61.74 | #31 |
| 4 | Claude Fable 5.1 | 61.53 | #10 |
| 5 | Claude Opus 5.5 | 61.33 | #3 |
| 6 | Gemini 3 Pro (Preview) | 60.3 | #71 |
| 7 | Claude Sonnet 5.5 | 60.18 | #14 |
| 8 | Qwen 3.5 Plus (02-15) | 60.17 | #233 |
| 9 | Gemini 3.5 Flash | 60.05 | #66 |
| 10 | Grok 4.6 | 59.71 | #29 |
| 11 | Gemini 3.6 Flash | 59.48 | #57 |
| 12 | Gemini 2.5 Pro (Preview 05-06) | 58.94 | #156 |
| 13 | Gemini 3 Flash (Preview) | 58.35 | #90 |
| 14 | Gemini 2.5 Pro | 57.98 | #159 |
| 15 | Claude Opus 4.8 | 57.77 | #39 |
No result here: #6 GPT-6 Astra.
Interactive version: theaggregate.ai/benchmark?slug=tinybird-ai-sql-benchmark-exactness · How It Works · Data refreshed daily, snapshot 2026-11-05.