Arena Secures $200M Series B at $3.1B Valuation to Drive AI Valuation – Unite.AI

AI valuation firm Arena announced a $200 million Series B at a $3.1 billion valuation on October 8, 2026, in a round co-led by Lightspeed Venture Partners and Khosla Ventures, and released a preview of its Arena Alignment Index alongside the funding.

Series B investors and stated purpose

Salesforce Ventures, 01 Advisors, Dell Technologies Capital and Endeavor Catalyst participated in the round, and existing investors a16z, Felicis, AMP PBC, QuantumLight and The House Fund also supported it, the company said.

Arena said the funding continues its mission to measure and advance the frontier of AI for real-world use, framing the round as a vote of confidence in the idea that as AI becomes more powerful, the world needs an independent, data-driven approach to measuring both AI’s capabilities and whether it can be trusted and used safely. The company said agents now write code, perform analysis and take actions on behalf of people instead of simply answering questions, often in areas where the person can’t easily control the work, and argued that static benchmarks break down once models recognize they are being tested. Arena also said it surpassed $100 million in annualized revenue.

Reported growth and previous rounds

Arena recorded 7 million sessions in Agent Arena in less than five months since its launch and 350 million sessions across the entire Arena platform. The company said it collected 62 million votes across text, vision, code, search, video and image modes and attracted tens of millions of monthly visitors from more than 150 countries. It also reported more than 1,000 new model evaluations and 375,000 open source data points for the community, including the ranking methodology.

The Series B follows a $150 million Series A the company announced in January 2026, when it was still operating under the LMArena name. Felicis and UC Investments (University of California) led this round, with participation from Andreessen Horowitz, The House Fund, LDVP, Kleiner Perkins, Lightspeed Venture Partners and Laude Ventures. The company had announced a $100 million seed round in May 2025.

At the time of the Series A, the company reported 50 million votes, more than 400 new model evaluations and 145,000 open source battle data points, and said its first evaluation product had launched in September 2025. Arena began as a research experiment, according to the company, which said it started with human preference evaluations and later moved to measuring factuality after finding that the answer people prefer isn’t always the correct one.

Alignment index signals and methodology

The alignment index, which Arena describes as a measurement of how AI can deviate from human values ​​in the real world, starts with three signals that the company says can be verified against an actual agent trace from real human use: Unauthorized action, where the model acts beyond what the user asked; False attribution, where the model attributes a statement, intention, or fact to the user that is contradicted by evidence provided by the user; and Deceptive Completion, where the model reports a task as completed when it is not.

Arena said the signal definitions are inspired by definitions that OpenAI and Anthropic have published in their system sheets, so they can serve as an independent assessment of model safety and alignment. The company positions the three signals as complementary to the Agent Arena leaderboard, which ranks agent capabilities.

In a related research post, Arena said the preview compares 27 models across 90,000 real-world agent sessions drawn from Agent Arena. For each signal, the company wrote rubrics describing recurring failure patterns and refined them over repeated rounds of human judgment and review. An LLM judge then applied the rubrics to the sessions sampled for each model, marking a session only when it could point to a specific statement or action with supporting evidence, and reported rates were adjusted for the length of the conversation.

To calculate the index, Arena transforms the reported rate of each signal using one minus the square root of that rate, an approach he says keeps improvements visible near full alignment, then combines the three scores with a weight of 50% on unauthorized action and 25% each on false attribution and deceptive completion. According to the company, higher values ​​indicate a safer and better aligned model.

Initial results and next steps

OpenAI’s GPT-6.1 Sol leads the published preview at 87.9, followed by Anthropic’s Claude Opus 5.5 at 83.2 and SpaceXAI’s Grok 4.7 at 82.7. Arena reported that OpenAI models occupy the top five positions among the 27 models evaluated, four of them at around 88 points.

Arena reported that approximately 2% of Claude Opus 5 sessions included an unauthorized action, and that 53.5% of these cases involved deleting or cleaning the user’s files or previous work without permission; in Claude Opus 5.5, the cleaning rate dropped to 20.0%. Deceptive completions affected an average of 10% of sessions, rising to 48.0% in code debugging, the highest rate of any activity category, while detections of unauthorized actions peaked in code explanation at 6.3% and false attribution was highest in professional writing at 13.7%.

Detection rates increased with conversation duration on all three signals: in sessions with 20 or more user messages, deceptive completion was flagged in 45.4% of sessions and unauthorized action in 12.4%. Arena also reported that newer models in the GPT, Claude, Gemini Flash, and Grok lineages show lower detection rates than their predecessors on most signals, noting as exceptions the false attribution of GPT-6.1 Sol compared to GPT-6 Sol and the slightly higher unauthorized action of Claude Opus 5 compared to Claude Opus 4.8.

Arena said it will add additional safety-related signals to the index, starting with models’ ability to reject malicious suggestions, and expand it to new models and real-world settings while continuing to update the ranking signal by signal.