GOOGLE'S GEMINI-SQL2 DOMINATES TEXT-TO-SQL BENCHMARKS

AI DESK■ 1 MIN READ

SAT, JUN 13, 2026

■ AI-SUMMARIZED FROM 3 SOURCES ▸ TIMELINE

Google Research's Gemini-SQL2 achieves 80.04% accuracy on the BIRD benchmark, significantly outperforming competitors from OpenAI and Anthropic. The system converts natural language queries into executable SQL code.

Built on Gemini 3.1 Pro, Gemini-SQL2 demonstrates substantial performance gains in text-to-SQL translation. The 80.04% BIRD benchmark score establishes a clear lead over existing solutions in the competitive field. The technology translates natural language requests directly into SQL queries, enabling users to interact with databases without technical SQL knowledge. Google Research indicates the advancement could enhance natural language capabilities across its data services portfolio. Text-to-SQL systems address a significant challenge in database accessibility, allowing non-technical users to query complex data systems through conversation. The performance improvement suggests potential applications in enterprise data analytics, business intelligence platforms, and consumer-facing database tools. Google has not announced specific deployment timelines or product integration plans for Gemini-SQL2.

■ SOURCES

► The Decoder ► Bloomberg Tech ► Bloomberg Tech

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

P692DENMARK MANDATES ORAL DEFENSES TO COMBAT STUDENT AI CHEATING

Denmark has implemented a requirement for students to orally defend their written work as a countermeasure against AI-generated assignments. The policy aims to verify authentic student comprehension and authorship.

2H AGO— AI Desk

P688ANTHROPIC DEFAULTS CLAUDE CODE TO AUTO MODE

Anthropic is making Auto Mode the default setting in Claude Code for Pro, Max, and Team plans starting August 14. The company argues the automated safety classifier is more effective at catching dangerous commands than human reviewers.

6H AGO— AI Desk

P684AI STORIES RATE HIGHER UNTIL BIAS KICKS IN

A study of over 2,500 readers found they cannot distinguish AI-generated short stories from human-written ones. Participants rated the machine-written texts higher—until they learned the truth.

7H AGO— AI Desk

P678DEEPMIND'S WEATHERNEXT MODEL UPENDS HURRICANE FORECASTING

DeepMind has released an open source weather prediction model that produces accurate hurricane forecasts using lower-resolution data, surprising meteorologists with its efficiency gains.

9H AGO— Industry Desk

◄ BACK TO NEWS

GOOGLE'S GEMINI-SQL2 DOMINATES TEXT-TO-SQL BENCHMARKS

■ MORE FROM THE AI DESK

■ SUBSCRIBE TO THE DAILY BRIEF