Vals AI, backed by a16z, pushes to become neutral referee for AI benchmarks
The startup keeps its tests secret and grades models on real industry tasks to stop firms from gaming benchmarks.
What to know
- Vals AI argues legacy AI benchmarks are outdated and easily gamed because their test materials are public, letting companies train against the exam.
- Vals instead keeps its test materials private and grades models on industry-specific real-world tasks in law, finance, and coding rather than abstract knowledge tests.
- The startup raised a $40 million Series A led by Andreessen Horowitz last month, following an earlier seed round from 8VC and Bloomberg Beta.
- Vals is expanding its evaluation scope into novel and high-stakes domains, including recursive self-improvement, biosecurity, cybersecurity, mental health, and law-of-armed-conflict compliance.
“Historically, I think evaluation has been done to evaluate intelligence in a very abstract way. Like, do models know enough information to be able to take a bar exam type test?”
Rayan Krishnan, Vals AI co-founder · TechCrunch ↗ · Sep 18
Rayan Krishnan 25-year-old co-founder of Vals AIVals AI AI benchmarking startup founded 2024Andreessen Horowitz Lead investor, Series A8VC and Bloomberg Beta Lead investors, seed round
How it unfolded 1 development · click the chart to see its coverage articlesposts
-
1
TechCrunch profiles Vals's private, task-based benchmarking approach
In an interview and office tour, Krishnan described Vals's method of testing models on undisclosed, industry-specific tasks in law, finance, and coding, and its expansion into recursive self-improvement, biosecurity, cybersecurity, mental health, and law-of-armed-conflict benchmarks.
“We have a benchmark on recursive self improvement. We're doing some work in mental health, cybersecurity, biosecurity, and even law of armed conflict to models to understand how to apply the Geneva Convention…”
— Rayan Krishnan -
1 outlet Vals, backed by Andreessen Horowitz, is looking to become the gold standard for AI benchmarking
first by TechCrunch, 7d ago
-
N
Vals just raised a 0M Series A led by a16z to become the gold standard for AI benchmarking. Private test sets, real-world task evals, federal-agency program. 8x revenue YoY.
2 more of the top 3 · 6 posts in this stretch
-
T
Vals AI is hoping to make AI benchmarking a more neutral and trustworthy resource in a world increasingly inundated by AI models. https:// techcrunch.com/2026/09/19/vals -backed-by-andreessen-horowitz-is-looking-to-become-the-gold-standard-for-ai-benchmarking/?utm_source=dlvr.it&utm_medium=mastodon
-
M
🚨 BREAKING: Vals is a startup that aims to improve AI benchmarking systems. It raised $40 million in a series A funding round led by Andreessen Horowitz in September 2026.
-
-
background
Vals raises $40M Series A led by Andreessen Horowitz — After a period of rapid growth, Vals closed a $40 million Series A round led by a16z, deepening its push into AI evaluation infrastructure.
-
background
Vals closes a seed round led by 8VC and Bloomberg Beta — The young company secured seed funding as it began establishing itself in the AI industry.
-
background
Rayan Krishnan co-founds Vals AI — Krishnan, then a Stanford undergraduate who had interned at Palantir and worked at Microsoft and Stanford's AI lab, started Vals after observing that academic benchmarks were not keeping pace with fast-advancing frontier models.
What people are saying 3 voices from 1 site · best of 6 · verbatim
- Sep 24
-
K
Vals, respaldado por Andreessen Horowitz, quiere convertirse en el estándar de oro para la evaluación de IA —
- Sep 19
-
D
Imagination sparks inside the machine: Vals, backed by Andreessen Horowitz, is looking to become the gold standard for AI benchmarking Source: techcrunch.com #techcrunch #ai
-
A
TechCrunch AI: AI Benchmarking and Evaluation Standards Emerge as Critical Infrastructure Link: https://techcrunch.com/2026/09/19/vals-backed-by-andreessen-horowitz-is-looking-to-become-the-gold-standard-for-ai-benchmarking/ | Context: