Benchmarking has become a standard practice for AI companies to measure and showcase the performance of their models. It allows them to compare their technologies with competitors and highlight their strengths. Vals, a startup founded in 2024, is challenging the current benchmarking system, which it claims is outdated and insufficient for evaluating the rapidly evolving AI landscape. In less than two years, Vals has emerged as a prominent player in the tech industry, securing a seed round led by 8VC and Bloomberg Beta last year. Recently, the company raised $40 million in a Series A round led by Andreessen Horowitz, signaling strong investor confidence in its mission. Rayan Krishnan, one of Vals’ 25-year-old co-founders, brings a strong background in AI and technology. He previously interned at Palantir and worked at Microsoft and Stanford’s AI lab. Krishnan founded Vals after noticing that traditional academic benchmarks were not keeping pace with the rapid advancements in AI models. “We saw new, highly capable models entering the market quickly, but the existing benchmarks weren’t reflecting that progress,” he explained. Vals differentiates itself by testing AI models on their ability to perform complex, industry-specific tasks, rather than relying on general knowledge tests. This approach aims to measure how well AI can perform real-world tasks in fields like law, finance, and coding. “We’re looking at the real-world impact of these models,” Krishnan said. “Can they produce work of the same quality as a human in every domain?” The scope of Vals’ evaluations is expanding beyond traditional industries. The startup is exploring unique areas such as recursive self-improvement, mental health, cybersecurity, and even the application of international law like the Geneva Convention. Krishnan mentioned that Vals is also working on benchmarks related to the law of armed conflict, helping models understand how to apply international regulations in complex scenarios. As Vals grows, so does its business. The startup’s revenue has increased eightfold compared to last year, and its team has expanded from eight to 25 employees. Krishnan plans to move to a larger office and hire an additional 10 to 15 people. The company has also launched a program to provide model evaluations to federal agencies. Krishnan believes that as AI becomes more integrated into the economy and more companies go public, benchmarking systems like Vals’ will become essential in building trust and guiding future investments in AI technology.