Founded in 2024, Vals keeps its benchmark test materials private so companies cannot train against them.
Vals raised $40 million in a Series A round led by Andreessen Horowitz. The startup formed in 2024 and previously raised a seed round led by 8VC and Bloomberg Beta. Vals builds AI benchmarks that it does not publish, an approach meant to stop model makers from training against the tests. Co-founder Rayan Krishnan, 25, started the company after watching academic benchmarks fall behind frontier models.
Benchmarking has become the standard way AI companies prove what their models can do, and strong scores double as marketing. Legacy tests were built for older systems and many are public, which lets a vendor train directly against the questions. Vals instead evaluates models on tasks tied to law, finance, and coding, and it also probes for negative outcomes.
For builders and operators, benchmark scores in model cards and vendor pitches rest on public tests that a vendor can game. A closed evaluation set gives buyers a cleaner signal when they compare models for a specific domain. Enterprises that pick a model for legal work or code generation still need internal evals. Vals aims to supply the outside check that procurement and platform teams lack.
Vals plans to keep its test materials private, so its value depends on whether enterprises and model makers accept its scores as a fair comparison. The company has not said which domains it will cover next or how it prices access. Watch whether independent labs reproduce Vals results and whether regulators or buyers cite them in procurement. The Series A funds expansion of those evaluation sets.
What matters
- Vals raised a $40 million Series A led by Andreessen Horowitz after a 2024 founding.
- Public benchmarks let vendors train against the test, so operators need evaluation they cannot game.
- Watch whether Vals publishes results that enterprises adopt when choosing models for law and finance.
Why it matters
Watch whether Vals publishes results that enterprises adopt when choosing models for law and finance.
This GenAI News article was prepared in original wording using reporting and materials published by TechCrunch AI. Source reference: https://techcrunch.com/2026/09/19/vals-backed-by-andreessen-horowitz-is-looking-to-become-the-gold-standard-for-ai-benchmarking/.
Drafted by the GenAI News review pipeline.