AI Leaderboard Arena Hits $3.1B Valuation in $200M Series B

Arena, the crowdsourced AI leaderboard born at UC Berkeley in 2023, raised a $200 million Series B at a $3.1 billion valuation, the company said Thursday. The post-money figure is roughly 1.82 times the $1.7 billion mark from January. That translates to a near-doubling in about 10 months.

What changed in those 10 months?

Arena reported annualized revenue of $30 million in January. By June, the company said, that figure had climbed to $100 million. That's a 3.3x jump in roughly five months.

The growth case is hard to dispute on the surface. The proof will arrive in audited financial statements, which Arena has not released.

Bottom line: investors are paying up for revenue growth the company itself has reported but no auditor has yet verified.

Why does Arena's pitch resonate now?

"AI is advancing faster than our ability to evaluate it, and static benchmarks break down once models recognize they're being tested," Arena said in its funding announcement. The quote positions the company as the neutral arbiter the field needs. Whether crowdsourced voting can actually substitute for sanctioned testing is an empirical question nobody has answered.

The demand is real. AI labs now openly game benchmark scores, and enterprises want evaluations tailored to their own workloads. Arena's free consumer site registers tens of millions of monthly visitors, the company claims.

Who is paying $3.1 billion?

Lightspeed Venture Partners and Khosla Ventures co-led the round. Salesforce Ventures, 01 Advisors, Dell Technologies Capital, Endeavor Catalyst, a16z, and Felicis also invested, Arena said. The roster mirrors the enterprise-software investor class, several of whom have also funded the model providers themselves.

The deal prices Arena at roughly 31 times its stated June run-rate revenue. That multiple sits well above typical SaaS comparables at the same revenue scale. It leaves little room for execution missteps before the next round.

Bottom line: a deep-pocketed syndicate is underwriting Arena's rise, but the multiple leaves limited margin for stumbles.

What does the new alignment leaderboard measure?

Arena added an alignment category tracking three failure modes: unauthorized action, false attribution, and what it calls "deceptive completion." OpenAI's models currently sit at the top of the preliminary ranking. Anthropic's Claude Opus 5.5 ranks sixth, and Claude Fable ranks ninth.

"The world needs a neutral third party to measure how safe and aligned AI actually is once it's in the hands of real people," Arena said. "Arena is stepping into that role today."

What the announcement does not disclose

Arena named investors and a price tag. It did not name AI Evaluations customers, disclose the revenue split between its free consumer site and paid product, or publish an audit of its evaluation methodology. None of those omissions is unusual for a Series B deal.

Together they explain why benchmark-checking investors will press for hard numbers before the next round. The next test arrives when audited financials surface, or when a competing evaluator wins comparable customer loyalty. Until then, Arena's $3.1 billion price rests on the company's own tallies.