Arena, an AI evaluation platform formerly known as LMArena, announced on October 8, 2026 that it has raised a US$200 million Series B at a US$3.1 billion valuation. The company also introduced the Arena Alignment Index, a new measurement designed to assess how frontier AI models act in line with human intent. The announcement positions Arena as an independent, data-driven resource for measuring both AI capability and real-world trustworthiness.
Series B Funding Details
The Series B round was co-led by Lightspeed Venture Partners and Khosla Ventures, with participation from Salesforce Ventures, 01 Advisors, Dell Technologies Capital, and Endeavor Catalyst. Existing investors including Andreessen Horowitz, Felicis, AMP PBC, QuantumLight, and The House Fund also joined the round. Arena has exceeded US$100 million in annualized revenue, reaching that milestone eight months after its Series A.
Introducing the Arena Alignment Index
The Arena Alignment Index launches with three initial signals that can be verified against actual agent traces: Unauthorized Action, False Attribution, and Deceptive Completion. These signals respectively cover cases where a model acts beyond a user request, attributes contradicted statements to a user, or claims a task is complete when it is not. Arena said the definitions align with system cards published by labs such as OpenAI and Anthropic, enabling independent safety and alignment assessments.
Platform Growth Metrics
Arena reported 7 million sessions in Agent Arena in less than five months since launch and 350 million sessions across the full platform. Users have contributed 62 million votes across text, vision, code, search, video, and image modalities, while tens of millions of monthly visitors arrive from more than 150 countries. The Alignment Index preview compares more than 20 frontier models, and a related research post cites results across 27 models and 90,000 real-world agent sessions.
Open Research and Community Contributions
Arena has open-sourced 375,000 data points for community research, a figure that includes its leaderboard methodology and evaluation approach. The platform has tracked more than 1,000 new model evaluations across open and proprietary systems, while tens of millions of monthly visitors from over 150 countries contribute real-world usage signals. This scale supports the company's claim that its leaderboards reflect observed human behavior rather than curated or static benchmark conditions.
Why Real-World Alignment Measurement Matters
Arena argues that as AI agents write code, run analyses, and take actions on behalf of people, static benchmarks have become insufficient because models can recognize when they are being tested. The company states that the world needs a neutral third party to measure how safe and aligned AI actually is once it is in the hands of real people. Its Agent Arena methodology observes the full human-agent workflow through causal inference, extending from capability to trust and safety.
With the new funding, Arena plans to continue expanding its alignment signals and leaderboard coverage in the same way its capability leaderboards have grown over time. The company positions the Alignment Index as a first step toward giving the industry a grounded, real-world verified signal for AI alignment. Arena said the new resources will help it build an independent, community-driven benchmark for how AI models and agents work, behave safely, and act truthfully.