Vals AI is an independent AI benchmarking and evaluation company based in San Francisco. It designs, runs, and publishes rigorous tests for large language models (LLMs) across professional domains: finance, law, tax, coding, cybersecurity, and now mental health. Rather than recycling academic toy problems, Vals builds domain-specific benchmarks that simulate real work — think Excel financial modeling, legal case research, or tax code reasoning — and publishes public leaderboards ranking every major frontier model head-to-head. Foundation model labs (OpenAI, Anthropic, Google, Meta, xAI) cite Vals results in official model cards. Enterprises use the platform to pick the right model for a given workflow. Hospital systems use it to assess AI deployments in clinical settings. The business has two sides: free public leaderboards that drive credibility and brand, and a paid private evaluation infrastructure sold to labs and enterprise engineering teams who need custom, confidential testing before deploying AI into high-stakes workflows.