Vals Aims to Revolutionize AI Benchmarking with New Approach
Vals, a startup founded in 2024 and backed by Andreessen Horowitz, is working to modernize the way artificial intelligence models are benchmarked. Traditional benchmarks have become outdated as companies find ways to manipulate them for better results. Vals aims to address this issue by developing more robust evaluations that test AI's ability to perform real-world tasks in industries like law, finance, and coding rather than just general knowledge tests. The company recently raised $40 million in a Series A funding round led by Andreessen Horowitz after securing an earlier seed investment from 8VC and Bloomberg Beta. Vals' co-founder Rayan Krishnan explains that the startup's approach focuses on measuring both positive outcomes and potential negative impacts of AI models, ensuring they can produce work comparable to humans across various domains. The company is expanding its benchmarking capabilities into areas such as cybersecurity, biosecurity, mental health, and even law of armed conflict. Vals' revenue has grown eightfold year-over-year, with the team now tripling in size since the beginning of 2024.
Topics: AI Benchmarking Funding Rounds Startups