Skip to content
TrustList
Funding news

Arena raises $200M Series B led by Lightspeed and Khosla

Editorial

By TrustList Editorial

The AI model leaderboard began as a UC Berkeley research project in 2023 and says it now logs tens of millions of monthly visitors from more than 150 countries.

About Arena raises $200M Series B led by Lightspeed and Khosla

Arena raises $200M Series B led by Lightspeed and Khosla

8 October 2026: Arena, the company behind the public leaderboard that ranks AI models by crowd vote, has raised a $200 million Series B led by Lightspeed Venture Partners and Khosla Ventures. TechCrunch reports the round values the company at $3.1 billion, close to double the $1.7 billion post-money valuation of its Series A in January.

Arena says the new investors include Salesforce Ventures, 01 Advisors, Dell Technologies Capital and Endeavor Catalyst. Existing backers a16z, Felicis, AMP PBC, QuantumLight and The House Fund also took part. The January Series A was $150 million.

The product is free to use: visitors compare answers from two anonymous models and vote for the better one, and those votes build the rankings. Arena says it has collected 62 million votes across text, vision, code, search, video and image tasks, and that its platform has run 350 million sessions. It also says its Agent Arena, launched less than five months ago, has logged 7 million sessions.

The revenue comes from a separate service. Arena sells AI Evaluations, a performance-analytics product for model labs and for enterprises that need to compare models before buying. The company says its annualized revenue is above $100 million. TechCrunch puts that figure at June.

The announcement names no individual speaker. Its unsigned statement, from the "Arena Team", says the round is a vote of confidence as AI becomes more powerful, and that Arena started as a research experiment and is becoming how the industry checks whether models and agents work.

One product detail drives the timing. Arena has added an Alignment Index that ranks models on behaviour such as taking actions nobody authorised and claiming to finish a task they did not complete. The company says the funding gives it the resources to build that index properly, and it says it has open-sourced 375,000 data points.

For teams choosing between models, a crowd-voted ranking measures general preference, not fitness for a particular workflow, so the commercial evaluation service is the part aimed at that gap. Our AI model category lists 112 products.

Arena does not say how it will split the money between the leaderboard, the enterprise service and the alignment work.

Company profile on TrustList: Arena

Sources

Categories & features

TrustList Weekly

The week in software and IT, in one email

The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.

We will email you to confirm. Unsubscribe with one click in any issue. Privacy policy