AI news · Funding · evaluation
Arena raises $200 million and ranks models on alignment
The leaderboard company is valued at $3.1 billion and its new index finds false task-complete claims in 48% of debugging sessions.
Arena, the crowdsourced AI leaderboard that began as a UC Berkeley research project, raised a $200 million Series B at a $3.1 billion valuation, it said on Thursday 8 October 2026. Lightspeed Venture Partners and Khosla Ventures co-led the round. Arena says its annualised revenue has passed $100 million.
Its Series A in January raised $150 million at a $1.7 billion post-money valuation, when revenue stood at $30 million, so the valuation has nearly doubled in about ten months, TechCrunch notes.
Arena also launched an Alignment Index that scores 27 models on 90,000 real agent sessions. Its three signals catch a model overstepping the request, pinning a claim on the user that the user’s own evidence disproves, and calling unfinished work finished. An AI judge applied written rubrics to a sample of sessions for each model. OpenAI models hold the top five places, four of them at about 88 points, while Claude Opus 5.5 and Grok 4.7 score 83. By Arena’s count, false completion claims affect 10% of sessions on average and 48% in code debugging. In sessions of 20 messages or more, about one in eight includes an unauthorised action.
For teams handing agents real work, the debugging figure argues for checking every “fixed” claim before it ships.
Sources
- Arena: Measuring the AI Frontier for Real-World Alignment: Arena’s $200 Million Series B. Checked
- Arena: Arena AI Alignment Index. Checked
- TechCrunch: Popular AI leaderboard Arena nearly doubles valuation to $3.1B valuation in 10 months. Checked
Alexandre S.
Alexandre S. is a creative ad strategist and copywriter with ten years' experience in digital marketing across multiple channels.