Arena, a leading AI leaderboard provider that began as a UC Berkeley research project in 2023, has achieved an impressive $100 million in annualized run-rate revenue, merely eight months after the launch of its commercial services.
The company is widely recognized for its highly popular crowdsourced AI model performance leaderboard, which is meticulously compiled from over 10 million user evaluations. Its consumer-facing website allows users to input a prompt, which is then sent to two distinct AI models. Users subsequently assess and select the model that delivered superior performance.
While Arena’s renowned AI model leaderboard remains freely accessible to the public, the company commenced its revenue generation in September with the introduction of AI Evaluations. This service offers model laboratories and enterprises comprehensive, deep-dive performance analytics, all derived from the insights of its engaged community.
Arena’s remarkable revenue acceleration underscores the strong appeal of its commercial offerings, which resonate as powerfully with its clientele as they do with its community of evaluators. This community is frequently drawn to the platform by the opportunity for early access to the latest, often unreleased, AI models.
“A lot of people don’t even understand that our business is making any money at all; people still see us as like an open-source project,” Anastasios Angelopoulos, Arena’s co-founder and CEO, shared with TechCrunch, highlighting a common misconception about the company's operational model.
Although Arena refers to its revenue milestone as ARR, a term traditionally signifying annualized recurring revenue, Angelopoulos clarified that the company invoices customers based on "consumption," indicating that its revenue model is not recurring in nature.
Despite the absence of direct competitors – notably, Yupp, another crowdsourced AI model-picking startup, ceased operations in March – Angelopoulos stated that Arena actively competes “for the same dollar” with human labeling startups such as Mercor, Surge, and Scale AI. These companies all provide essential services to model developers for refining their AI during the crucial post-training phase.
The demand for post-training refinement services continues to escalate as AI providers persistently strive to maximize model performance. When Arena announced its $150 million Series A funding round in January, which valued the company at $1.7 billion post-money, its annualized revenue stood at $30 million.
In a broader market context, Handshake’s gross annualized revenue from AI training has almost doubled since January, soaring from $550 million to nearly $1 billion, as reported by The Information in April. Similarly, Mercor’s annualized revenue also surpassed $1 billion earlier this year, an increase from $500 million last September, according to The Information.
Arena’s evaluation capabilities span a diverse array of tasks, including text processing, coding, vision, and image generation. Furthermore, it supports complex, long-running workflows through its recently launched Agent Mode.
Arena was co-founded by Anastasios Angelopoulos, alongside fellow UC Berkeley postdoctoral student Wei-Lin Chiang, who serves as the startup’s CTO. The founding team also includes Ion Stoica, a distinguished UC Berkeley professor and co-founder of Databricks, who provided advisory support to the project before its incorporation as a company, noted as April 2025.
To date, Arena has successfully raised a total of $250 million from a notable roster of investors, including Felicis, Andreessen Horowitz, The House Fund, LDVP, Kleiner Perkins, Lightspeed Venture Partners, Laude Ventures, and UC Investments.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.