Vals Raises $40 Million Series A to Develop New AI Benchmarking Standards

Funding for AI Benchmarking

Vals, an AI startup established in 2024, announced it has raised $40 million in a Series A funding round. Andreessen Horowitz led the investment, following a seed round previously led by 8VC and Bloomberg Beta. This funding will support Vals' mission to improve AI model evaluation.

Addressing Benchmarking Limitations

The company was formed to address shortcomings in existing AI benchmarking systems. Co-founder Rayan Krishnan noted that many legacy benchmarks are not designed for modern AI models, leading to situations where companies can optimize models specifically for these tests. Vals aims to provide more robust and relevant evaluations.

New Evaluation Approach

Vals differentiates its approach by not publicly disclosing its specific test materials, which prevents models from being trained directly against the benchmarks. Instead of focusing on abstract intelligence tests, Vals evaluates AI models on their ability to complete complex tasks. This method seeks to verify what models can actually accomplish, rather than just their general knowledge.

Industry Impact

The development of new, more secure, and task-oriented AI benchmarking could influence how AI capabilities are validated and advertised across the industry. As AI integration expands, accurate and reliable evaluation methods become more critical for verifying model performance claims.