Genloop Tops LiveSQLBench at 68.15%, Outperforming the Best AI Agents from OpenAI and Anthropic

Independent benchmark results put Genloop in a class of its own for enterprise SQL and data reasoning - while leaving source data untouched.
Santa Monica, CA - June 2, 2026 - Genloop [https://genloop.ai/], a portfolio company of 1752vc [https://www.1752.vc/], today announced that its platform has scored 68.15% on LiveSQLBench, the contamination-resistant benchmark for live, real-world database reasoning - a result that the best agents from OpenAI and Anthropic have not been able to match, with neither passing the 50% mark.

On read-only queries, Genloop reached 90.5% accuracy. Just as important as the score is how it was achieved: Genloop does not manipulate source data to arrive at an answer. The system reasons over enterprise databases exactly as they are, preserving data integrity in the kinds of high-stakes environments where an incorrect or destructive query can carry real financial consequences.

The LiveSQLBench [https://livesqlbench.ai/] result extends a pattern Genloop has already established elsewhere. The company ranks #1 globally on Spider2 [https://spider2-sql.github.io/]- widely regarded as the most rigorous benchmark for complex SQL reasoning and multi-step investigation on real enterprise data - with a score of 96.7, ahead of leading technology organizations including Tencent, AT&T, ByteDance, and Snowflake. On LiveSQLBench, the story is the same: Genloop leads.

These benchmarks matter because enterprise data is messy, fragmented, and risky - long considered too unreliable for dependable AI reasoning. Genloop was built to prove otherwise. At the core of its platform is a proprietary LLM customization stack that powers AI agents acting as a personal data analyst for every team. These agents securely connect to enterprise data systems and translate complex datasets into trustworthy, immediately actionable insights, combining advanced reasoning, contextual memory, and enterprise-grade security and compliance.

Genloop is building exactly the kind of AI we believe will define the next generation of enterprise software. Clearing a benchmark like LiveSQLBench while the strongest general-purpose agents stay under 50% is precisely why we backed this team, and we're proud to support them through our Accelerate program," said a 1752vc spokesperson.

Genloop is a portfolio company of 1752vc and is participating in the firm's Accelerate program, which pairs a $100,000 investment with hands-on go-to-market and sales support. Founded by a team from Stanford, the IITs, and leading AI organizations, Genloop is building safe and sovereign AI systems designed for mission-critical enterprise environments.

About Genloop

Genloop is reimagining business intelligence for the GenAI era. Its proprietary LLM customization stack powers AI agents that act as personal data analysts - delivering instant, accurate insights with enterprise-grade security and compliance. Genloop is the most accurate data reasoning platform, ranking #1 on Spider2 and leading on LiveSQLBench. Learn more at genloop.ai.

About 1752vc

1752vc (formerly Pegasus Angel Accelerator) is the #1 investment and growth acceleration firm in Southern California dedicated to identifying and scaling high-potential startups. Combining strategic capital with deep industry expertise, 1752vc helps transform innovative businesses into market leaders. Headquartered in Santa Monica, the firm's team of seasoned entrepreneurs, operators, and investors brings decades of experience building and scaling successful companies.

Learn more at 1752.vc and http://1752.ai.

Media Contact
Company Name: VentureVoice Media
Contact Person: Mila Hart
Email:Send Email [https://www.abnewswire.com/email_contact_us.php?pr=genloop-tops-livesqlbench-at-6815-outperforming-the-best-ai-agents-from-openai-and-anthropic]
Country: United States
Website: http://venturevoicemedia.com

Legal Disclaimer: Information contained on this page is provided by an independent third-party content provider. ABNewswire makes no warranties or responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you are affiliated with this article or have any complaints or copyright issues related to this article and would like it to be removed, please contact retract@swscontact.com



This release was published on openPR.