Vals AI Lands $40 Million Series A to Scale Enterprise Model Benchmarking

Vals AI closed a $40 million Series A this week, led by Andreessen Horowitz. At an estimated $1.3 million in annual recurring revenue as of late 2025, a16z is underwriting the round at roughly 30 times trailing ARR — a multiple reserved for category leaders, not early-stage startups.

The valuation math

The cap table math is the headline. A $40 million Series A on $1.3M ARR prices Vals AI at a multiple that only works if the company expands its product line aggressively within 18 months. The SaaS News confirms the round; a16z is the only named investor on the cap table so far, which leaves the question of how much dry powder remains for a potential Series B when burn accelerates. Domain expert hiring — the stated use of proceeds — is a fixed cost that compounds monthly. Vals AI's path to justifying this valuation hinges on converting the Vals Index from a public benchmark into a paid continuous-monitoring product, not a marketing dashboard.

The credibility gap

Vals AI is selling a verdict on models that OpenAI, Anthropic, Google, and Meta all claim to lead. The pitch: enterprise buyers can't tell which vendor is winning when every lab publishes self-serving benchmark results. Vals Index sidesteps academic metrics — logic puzzles, fill-in-the-blank tests — in favor of task-specific scoring across financial analysis, coding, legal research, and web search. The August 2026 update puts "Claude Fable 5" at 75.14%, a single number that does more for procurement decisions than most white papers. Andreessen's bet is that this kind of neutral ranking becomes infrastructure the way Gartner or G2 once did for enterprise software — a toll booth between vendors and buyers.

Reality check

The sobering question: benchmarks are cyclical. When the next model generation ships, the leaderboard flips, and the index's perceived authority resets. Vals AI's stated plan — letting enterprises run evaluations on proprietary data — is the right defensive move, but it competes with in-house eval teams at the same Fortune 500 buyers who fund a16z's other portfolio companies. The next 12 months will show whether $40M bought a benchmark or a business. Right now, it's priced like the latter.