AI Evaluation Startup Vals Raised $40 Million Series A
The San Francisco-based firm plans to expand its model testing services following an eight-fold increase in revenue.
Updated on Sept. 19, 2026 in Startups
Live Poll
Should AI companies be required to use independent benchmarking to prove their performance claims?
In August 2026, the San Francisco startup Vals closed a $40 million Series A funding round led by Andreessen Horowitz. Founded in 2024, the company offers model evaluation services to identify performance weaknesses in AI systems across law, finance, and coding.
Why it matters
Vals aims to bridge the gap left by academic benchmarks that struggle to keep pace with rapid AI advancements. The company now provides testing services to federal agencies to help verify model capabilities and measure potential negative impacts.
The firm reported revenue growth of 8x compared to last year alongside a staff increase from eight to 25 employees. It is currently evaluating AI models based on their performance in complex tasks within the legal, financial, and programming domains.
The players
Vals
A San Francisco-based startup that provides performance evaluation services for AI models in the law, finance, and coding sectors.
Andreessen Horowitz
A prominent venture capital firm known for its large-scale investments in enterprise software, AI, and crypto-native startups.
The details
Vals evaluates artificial intelligence models by putting them through performance-testing tasks to identify operational weaknesses. Companies hire the startup to stress-test their models in specialized fields like law, finance, and software engineering. While these methods are designed to assess accuracy and reliability, the company maintains proprietary control over the specific evaluation datasets and test materials used in its audit processes.
Timeline
2024: The company was formed in San Francisco.
2025: Vals secured its seed funding round.
August 2026: The startup raised $40 million in a series A funding round.
September 2026: The company leadership provided a tour of their Folsom Street headquarters.
The Tech Race
As standard academic benchmarks for AI become increasingly saturated, specialized evaluation firms are competing to offer more rigorous, domain-specific testing. Vals is positioning itself as a verification layer for enterprises and federal agencies that require deeper metrics than public leaderboards currently provide.
The startup plans to move to a larger office and hire up to 15 additional employees, signaling an expansion of its testing capacity. While the service is currently focused on enterprise and federal clients, the underlying methodologies are likely to influence the reliability and safety standards of the AI tools appearing in consumer workflows.
The takeaway
The rapid growth of companies like Vals highlights the increasing demand for external validation in an era of opaque AI development. Industry observers should watch for the company's next hiring phase and any potential move to a larger headquarters as indicators of its scaling capability.
Further reading
For more on the local ecosystem of emerging technical firms, visit Startups.
Live Poll
Should AI companies be required to use independent benchmarking to prove their performance claims?
