SF-based Vals, which develops evaluations and benchmarks to test AI models on real-world tasks, raised a $40M Series A led by a16z at a $400M valuation
- Vals AI raised $40 million in a Series A round at a $400 million valuation. — Revenue has grown eightfold since 2025 …
Tech Funding NewsAbhinaya Prabhu
Context & Ripple Effects
Vals’ round brings venture backing to the measurement layer of AI deployment: its product tests models on real-world tasks, while the company says revenue has grown eightfold since 2025. a16z has also backed AI application companies, including 11x’s earlier Series B.
The financing arrives amid investment in adjacent AI operating layers, from a marketplace for AI computing capacity to model-hosting platforms. Vals is differentiated in the supplied coverage by focusing on how models are evaluated rather than where they run.
First-order effects
Vals gains $40 million to expand its evaluations and benchmarks, with a16z becoming the lead investor at a $400 million valuation.
The reported eightfold revenue growth gives Vals a commercial signal to pair with its benchmark-focused product as it sells into AI-model deployment workflows.
Second-order effects
AI teams assessing models on real-world tasks have a better-capitalized specialist supplier, raising the importance of evaluation tooling alongside model hosting and compute access.
a16z’s backing links Vals to an investment portfolio that already includes AI sales automation, increasing investor attention on the tooling needed to measure whether deployed AI performs useful work.
Third-order effects
If funding continues to spread from compute and model hosting into measurement, AI infrastructure will be defined increasingly by the cost and reliability of a useful task, not solely by access to models or capacity.
The pattern points toward a more specialized AI stack in which independent evaluation providers can become a distinct control point between model builders and enterprise deployment.
The trend: AI investment is broadening from supplying models and compute to proving performance on the real-world tasks those systems are meant to automate.
We are excited to partner with the @ValsAI team! Vals test models on the work people actually want them to do with differentiated benchmarks and will continue to define the frontier
Recursive self-improvement is the next frontier in AI We're thrilled to be partnering with Vals on their RSI index and excited to show an early preview of the work we've been doing together. Stay tuned for more!
Congrats to Vals team and especially former HRT intern @langstonnashold on their raise! It's under discussed but an essential part of quant is just measuring and understanding the current situation, let alone saying something about the future. Vals is the “rating agency” for AI
Super pumped to partner with Vals AI team on this journey to build the ratings agency for the AI era. @RaghuRaghuram @stuffyokodraws @shangdaxu and the entire AI Infra team are very excited to be working with the Vals team!!
1/ We're thrilled to announce our investment in @ValsAI. AI has gotten very good at taking tests and winning trophies. However, it's still hard to prove it can actually do the work. That shift creates an enormous need for a trusted, independent evaluation layer.
We backed @RayanKrishnan and @langstonnashold of @ValsAI (PearX S23) out of their dorm room! Congrats on raising $40M series A from @a16z @8vc and of course @pearvc !🚀
Today we're announcing our $40M Series A at a $400M valuation, led by @a16z , with participation from existing investors @8vc, @pearvc, and @BloombergBeta and new investors @HRTVentures and @nextladder. Alongside the fundraise, three more announcements: - Vals Smith: is now [imag…
Boom! It has been a blast being a part of this journey for the past year. @RayanKrishnan and @langstonnashold are incredibly hardworking and talented people and it has been a pleasure to work with them and everyone else in the @ValsAI team. With everyone attempting to build one
The best way to hold the AI industry accountable AND for the AI industry to thrive is to have an independent, trusted evaluator. Excited to see @langstonnashold @RayanKrishnan and the whole @valsai team play that vital role...
We're thrilled to invest in Vals. A frontier model can look brilliant on a leaderboard and still struggle with the messy work that actually matters in the real world. @ValsAI takes a fundamentally different approach to evaluation: test models on the work people actually want
Congrats to @ValsAI on their $40M Series A! Proud to have first partnered with @RayanKrishnan and @langstonnashold at pre-seed when they joined PearX S23 and even more excited to double down on what's ahead. Let's go! 🚀🚀🚀
In early 2023, two Stanford students pitched an idea built around one question: what can AI models actually do? ChatGPT had just launched and the word “evals” did not exist. It seemed highly implausible that anyone would use, let alone pay for them. We backed @RayanKrishnan and
Almost seven years ago, @RayanKrishnan and I met on a pre-orientation backpacking trip. He would end up being one of the most intelligent people I would meet in undergrad. Post-graduation, I was going to take an offer at HRT, and he was preparing to apply to PhD programs.
Excited to be part of this. People often invoke Goodhart's Law to argue that evals lose their value once created. But if we want to automate every economically valuable job, like the AI labs claim, we're going to have to evaluate every economically valuable job. From ARC-AGI
Congrats @RayanKrishnan and the @ValsAI team on the huge raise! In the last year at Cognition we've already seen frontier benchmarks expire with each new model release. As the models continue to push the frontier of capabilities further out, new benchmarks are needed to keep