Vals Secures $40M Series A to Redefine AI Benchmarking

AI evaluation startup Vals has secured a forty million dollar Series A financing led by Andreessen Horowitz, accelerating its mission to redefine industry standards for artificial intelligence benchmarking. Founded in 2024 and headquartered in San Francisco, the company emerged from co-founder Rayan Krishnan assessment that legacy academic metrics are failing to keep pace with modern model capabilities. Krishnan, a twenty-five-year-old Stanford alum with prior experience at Palantir and Microsoft, noted that traditional evaluations often rely on publicly available datasets that modern models can easily memorize or game, rendering results less indicative of actual utility. To address these shortcomings, Vals employs a proprietary testing framework that remains undisclosed to the public, effectively preventing models from overfitting to known benchmarks. Rather than measuring abstract knowledge retention, the startup evaluates AI systems against complex, domain specific tasks across law, finance, software development, and healthcare. The evaluation methodology extends beyond performance verification to assess potential negative externalities, including vulnerabilities in cybersecurity, biosecurity risks, mental health handling, and compliance with international frameworks such as the Geneva Convention in armed conflict scenarios. This dual focus on capability and risk mitigation positions Vals as a critical infrastructure provider for responsible AI deployment. The company commercial traction has been rapid. Vals reported revenue growth of eight times year over year and expanded its workforce from eight to twenty five employees since the beginning of the year. The organization has also initiated a specialized program to deliver model assessments to federal agencies, underscoring the growing demand for standardized, third party validation across both private and public sectors. Krishnan compared Vals revenue model to standardized testing platforms, noting that enterprises increasingly treat rigorous evaluation as a necessary prerequisite for product refinement and competitive differentiation. Industry observers view standardized benchmarking as an emerging necessity for AI commercialization. As major artificial intelligence developers prepare for public market listings and seek to establish investor confidence, independent performance verification will likely become embedded in regulatory filings and procurement decisions. Vals is currently planning a relocation to a larger facility and intends to add ten to fifteen additional staff members to support its expanding evaluation infrastructure. By shifting focus from theoretical intelligence tests to measurable, real world operational impact, the startup aims to establish the benchmarking protocols that will underpin trust, safety, and market confidence in the next generation of enterprise AI systems.