Tinybird AI SQL Benchmark - Success Rate: 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. Success rate is the percentage of questions answered with SQL that executes, allowing up to three retries that feed the SQL API error back to the model; higher is better.
Metric: Questions answered with a valid query within 3 retries (%). Source: llm-benchmark.tinybird.live. Saturation forecast: Estimated already saturated. 405 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 100 | #357 |
| 2 | Gemini 2.5 Pro | 100 | #159 |
| 3 | GPT-5 | 100 | #105 |
| 4 | Gemini 2.5 Flash | 100 | #255 |
| 5 | Claude Sonnet 4.5 | 100 | #148 |
| 6 | Claude Sonnet 4.6 | 100 | #94 |
| 7 | GPT-5.4 | 100 | #83 |
| 8 | GPT-4.1 | 100 | #258 |
| 9 | GPT-5.5 | 100 | #33 |
| 10 | GPT-4o Mini | 100 | #598 |
| 11 | Claude Opus 4.6 | 100 | #67 |
| 12 | Claude Haiku 4.5 | 100 | #287 |
| 13 | Kimi K2.5 | 100 | #152 |
| 14 | DeepSeek V4 Pro | 100 | #112 |
| 15 | GPT-5 Nano | 100 | #439 |
No result here: #6 GPT-6 Astra.
Interactive version: theaggregate.ai/benchmark?slug=tinybird-ai-sql-benchmark-success-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.