IndicDB (Zero-Shot without Evidence) - Hinglish: leaderboard
IndicDB (2026) is a multilingual Text-to-SQL benchmark on relational databases built from Indian government open data, with the same tasks and gold SQL asked in English, Hindi, Bengali, Marathi, Tamil, Telugu and Hinglish. This board is execution accuracy in percent of the generated SQL on the Hinglish questions, using zero-shot prompting without the external evidence; higher is better.
Metric: Execution accuracy (%) of generated SQL on the Hinglish variant of the IndicDB Text-to-SQL tasks (20 databases built from Indian government open data, join depth up to six), zero-shot prompting without the external evidence; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V3.2 | 60.53 | #419 |
| 2 | MiniMax-M2.7 | 50.21 | #539 |
| 3 | Llama 3.3 70B Instruct | 43.13 | #1385 |
| 4 | Qwen 3 8B | 38.58 | #1873 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=indicdb-zero-shot-without-evidence-hinglish · How It Works · Data refreshed daily, snapshot 2026-11-21.