a16z Leads $40M Series A for AI Model Evaluation Firm Vals AI

According to data from Beating, AI model evaluation company Vals AI has completed a $40 million Series A round at a $400 million valuation, led by a16z. Existing investors including 8VC, Pear VC, and Bloomberg Beta also participated. Vals specializes in third-party evaluation of large models. Currently, many models report performance metrics measured and disclosed by the manufacturers themselves. Vals, however, conducts standardized evaluations across real-world tasks in programming, finance, law, and healthcare to compare different models objectively. The company states that its results have been cited in model cards from OpenAI, Anthropic, Google, Meta, and xAI, and its revenue this year has already reached eight times its projected full-year total for 2025. a16z is particularly attracted to Vals’ value as an independent evaluator. Public benchmarks are increasingly susceptible to score manipulation, data contamination, and targeted optimization by vendors—high rankings no longer guarantee real-world usability. Vals has also fully launched Vals Smith, enabling users to turn any GitHub repository into a coding benchmark. The system extracts real development tasks from historical pull requests and evaluates whether models can successfully complete them using hidden tests. Enterprises can now directly determine which model best suits their codebase, rather than relying solely on public leaderboards.