Google's Gemini-SQL2 achieves 80.04% accuracy on text-to-SQL translation, outperforming OpenAI and Anthropic competitors and potentially improving natural language features across Google's data services.

What happened: Google Research unveiled Gemini-SQL2, a text-to-SQL system built on Gemini 3.1 Pro that translates natural language into executable SQL database queries. On the BIRD benchmark, Gemini-SQL2 achieves 80.04 percent execution accuracy, placing it first; OpenAI's GPT-5.5-xhigh scores about 72.8 percent, and Anthropic's Claude Opus 4.6 lands around 70.9 percent.

Why it matters: Converting natural language to correct SQL is especially challenging because data is often layered and queries must account for complex business logic. Better SQL understanding could improve natural language features across Google's data services more broadly, according to Google.

What to watch: Google Research has not announced a public release of the model or published a research paper yet, so availability and technical details remain forthcoming.