Vals Comes Out of Stealth With a$22M Round, Chasing the AI Benchmarking Throne
Vals is making a loud entry into the AI evaluation space, and it’s not pulling any punches. The startup just landed $22 million in backing from Andreessen Horowitz and other notable investors, and their mission is pretty straightforward: become the neutral, trusted yardstick for measuring AI system performance. Look, the industry is absolutely flooded with companies claiming their model is the best, and honestly? Nobody really trusts those self-serving scores anymore. When you’re trying to figure out what is AI and how different systems stack up, you need a source that isn’t selling something — and that’s exactly the hole Vals is aiming to fill.
The timing here is kinda perfect, if you think about it. The number of available AI models has exploded recently, with new ones popping up weekly from everyone and their competitor. Developers and enterprises alike are struggling to keep pace, let alone make informed decisions about which models actually deliver results. Most benchmarks today come from the same companies selling the products, which raises eyebrows for real. Vals wants to change that playbook by offering independent, standardized evaluations that don’t have a vested interest in the outcome.
So what does this mean for the average person or business? Well, it means a path toward clearer answers when trying to understand things like AI Tokens, model efficiency, cost per output, and actual capability comparisons. No more guessing whether a model is really better or just better at marketing. The team behind Vals includes folks who’ve spent years working inside the AI industry, so they know where the body counts are buried. They’re betting that trust is the scarcest resource in AI right now — and honestly, they might be onto something.
Why It Matters
- Benchmark trust is broken. Right now, almost every major AI benchmark is run by the companies selling the models themselves. That’s like letting the exam writer grade their own students. Independent, credible evaluation is basically nonexistent, and the industry knows it.
- The market is saturated and confusing. With new AI models launching constantly, decision-makers are flying blind. Enterprises are spending millions on tools they can’t reliably compare, and that waste is only growing.
- Andreessen Horowitz is betting big on infrastructure, not just apps. This funding signals that top-tier VCs see benchmarking and evaluation as critical plumbing for the entire AI ecosystem — not some nice-to-have side project. If Vals pulls this off, it could reshape how everyone measures progress in the field.