Vals, backed by Andreessen Horowitz, is looking to become the gold standard for AI benchmarking
News Source
β’Sat, 19 Sep 2026 13:00:00 +0000
π° What Happened
A startup called Vals wants to become the leading tester of AI models. Vals was founded in 2024 and last month raised $40 million in a funding round led by Andreessen Horowitz. An earlier seed round was led by 8VC and Bloomberg Beta.
Benchmarks are standard tests that show what an AI model can do, and companies use good scores in marketing. The problem is that many older tests are public, so firms can train models on the answers.
Vals keeps its test material secret. Instead of general knowledge quizzes, it sets real tasks from fields like law, finance and coding. Its 25-year-old co-founder, Rayan Krishnan, says the aim is to check whether models work as well as people.
π The Backstory
Rayan Krishnan studied at Stanford and worked at a school AI lab, and interned at Microsoft and Palantir. He said he started Vals because academic tests were falling behind fast-moving models.
AI is now used in many parts of daily life. Buyers want proof that a model can really do what its maker claims. Independent testing has become a big business.
π― Why It Matters
When model scores are honest, buyers can pick the right tool and prices stay fair. If tests can be gamed, companies may pay for AI that does not work as promised.
A startup called Vals wants to become the leading tester of AI models. Vals was founded in 2024 and last month raised $40 million in a funding round led by Andreessen Horowitz. An earlier seed round was led by 8VC and Bloomberg Beta.
Benchmarks are standard tests that show what an AI model can do, and companies use good scores in marketing. The problem is that many older tests are public, so firms can train models on the answers.
Vals keeps its test material secret. Instead of general knowledge quizzes, it sets real tasks from fields like law, finance and coding. Its 25-year-old co-founder, Rayan Krishnan, says the aim is to check whether models work as well as people.
Rayan Krishnan studied at Stanford and worked at a school AI lab, and interned at Microsoft and Palantir. He said he started Vals because academic tests were falling behind fast-moving models.
AI is now used in many parts of daily life. Buyers want proof that a model can really do what its maker claims. Independent testing has become a big business.
When model scores are honest, buyers can pick the right tool and prices stay fair. If tests can be gamed, companies may pay for AI that does not work as promised.