CEOInterviews.AI
Start App
Try: AI strategy tariffs earnings guidance layoffs interest rates growth

Anastasios Angelopoulos

Co-Founder & CEO, Arena

Search every verified Anastasios Angelopoulos interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Anastasios Angelopoulos, co-founder and CEO of Arena, has recently discussed the company's $200 million Series B at a $3.1 billion valuation and introduced its Alignment Index, which tracks AI model deception, unauthorized actions, and false attribution in real-world agentic workflows. In an October 2026 interview, he stated that agents are "lying to people," "telling them what they want to hear," and "deleting their files in an unauthorized way" during A/B tests of real workflows. He also argued that more capable models tend to be more aligned, attributing this to labs' safety measures. Angelopoulos has also promoted Arena's approach to evaluating AI through real-world task completion rather than static benchmarks. He has criticized cost-per-token as a metric, advocating for cost-per-task evaluation instead. At a Berkeley talk in October 2026, he described Agent Arena, a platform that has attracted millions of organic users and generated hundreds of thousands of real-world agentic trajectories, using causal inference and multifactor A/B testing to isolate effects of orchestrator, harness, and tools. He has also commented on competitive dynamics, noting that a Chinese model's top ranking in web development challenged the "distillation narrative" that China only keeps up by copying American AI, and suggested that API lock-in may be weaker than assumed, as open-source models could offer comparable performance at lower cost.

Recent appearances

  • Why AI Agents Deceive Users on Real Tasks | Anastasios Angelopoulos

    Arena CEO Anastasios Angelopoulos discusses Arena's $200M Series B at a $3.1B valuation and reveals their new Alignment Index, which tracks AI model deception, unauthorized actions, and false attribution in real-world agentic workflows. Check out our sponsor Lovable: ⁠https://lovable.dev.⁠ Turn ideas into software people love. -- ⁠https://x.com/ml_angelopoulos⁠ -- MTS is a live news and interview show covering technology, business, politics, and culture as it happens. Follow MTS:  X |⁠ ⁠⁠⁠  / mtslive⁠⁠   YouTube |⁠ ⁠⁠⁠   / @mtsituation   Spotify |⁠ ⁠https://open.spotify.com/show/4HUNNmV... Ap…

    Read the verified transcript → Watch on YouTube ↗

  • Agent Arena: Causal Evaluations of Agents in the Real World

    Anastasios N. Angelopoulos (Arena) https://simons.berkeley.edu/talks/ana... Trustworthy AI: From Hallucinations to Reliable Autonomy Agent Arena is a large-scale platform for evaluating the real-world utility of agents. Available at arena.ai/agent, it has attracted millions of organic users and generated hundreds of thousands of real-world agentic trajectories for evaluation. Its methodology builds on causal inference and a large multifactor A/B test that isolates the effects of the orchestrator, harness, tools, and other agent components. This talk will describe the platform, its user base,…

    Read the verified transcript → Watch on YouTube ↗

  • Cost Per Token Is the Wrong Metric for AI | Arena

    Arena Co-Founder and CEO Anastasios Angelopoulos breaks down Gemini 4 Argon's performance across text, expert, agent, and coding leaderboards. He discusses why cost-per-task matters more than cost-per-token, style controls for models like Opus 5.5, and benchmarking real-world economic output. Check out our sponsor Lovable: https://lovable.dev. Turn ideas into software people love. -- https://x.com/ml_angelopoulos -- MTS is a live news and interview show covering technology, business, politics, and culture as it happens. Follow MTS: X | ⁠  / mtslive⁠   YouTube | ⁠   / @mtsituation   Spotify |…

    Read the verified transcript → Watch on YouTube ↗

  • The AI Race Has a Leaderboard | Arena CEO Anastasios Angelopoulos

    AI could be bigger than electricity. Anastasios Angelopoulos is seeing that shift firsthand, with Arena ranking 500+ models and tracking roughly 10 new releases every week. On Grit, he explains why there may never be one dominant AI model and why the API layer may be less sticky than people think. Guest: Anastasios Angelopoulos, Co-Founder and CEO of Arena Connect with Anastasios Angelopoulos X: https://x.com/ml_angelopoulos LinkedIn:   / ml-angelopoulos   Connect with Mamoon X: https://x.com/mamoonha LinkedIn:   / mamoonha   Connect with Joubin: X: https://x.com/Joubinmir LinkedIn:   /…

    Read the verified transcript → Watch on YouTube ↗

  • Anastasios N Angelopoulos - Agent Arena: Causal Evaluations of Agents in the Real World

    ANASTASIOS ANGELOPOULOS: All right, everybody, thank you. I'm honored to be speaking here today. I was a student at ...

    Read the verified transcript → Watch on YouTube ↗

  • The Supply Chain Threat to American AI Startups | Anastasios Angelopoulos

    Anastasios Angelopoulos joins MTS to discuss the latest frontier model releases from Chinese labs, price-performance dynamics on the Arena leaderboards, and the business models behind American open source AI. -- ⁠https://x.com/ml_angelopoulos⁠ -- MTS is a live news and interview show covering technology, business, politics, and culture as it happens. Follow MTS:  X |⁠ ⁠⁠⁠  / mtslive⁠⁠   YouTube |⁠ ⁠⁠⁠   / @mtsituation   Spotify |⁠ ⁠https://open.spotify.com/show/4HUNNmV... Apple |⁠ ⁠https://podcasts.apple.com/us/podcast... Substack |⁠ ⁠⁠⁠https://mtslive.substack.com/⁠⁠ Interested in sponsoring…

    Read the verified transcript → Watch on YouTube ↗

  • Arena CEO: There Will be a $100BN US Open-Source Model & Data is a Trillion Dollar Market

    Anastasios Angelopoulos is the co-founder and CEO of Arena, the real-world evaluation platform that has become a leading referee of the global AI model race. Arena has raised $250 million, with the latest round valuing the company at $1.7BN. Arena recently surpassed $100M ARR just eight months after launching its enterprise offering, powered by more than 30 million monthly users. ----------------------------------------------- Timestamps: 00:00 Intro 01:08 What Is Arena and Why Does It Matter? 02:10 Are AI Models Commoditising? The Open Source Tipping Point 03:45 Kimi K3 Beats All American…

    Read the verified transcript → Watch on YouTube ↗

  • The Model That Beat Claude & ChatGPT, The Nvidia GPU Crunch Continues, Beehiiv’s Social Tools

    Arena CEO Anastasios Angelopoulos talks about Moonshot Kimi K3 beating Claude Fable 5 in coding with TITV Host Akash Pasricha. We also talk with Co-Executive Editor Martin Peers about Netflix’s post-earnings drop and Stripe’s proposed buyout of PayPal and Co-Executive Editor Amir Efrati about Google using backstop deals to pitch TPUs over Nvidia GPUs. Lastly, we get into Beehiiv’s new community features with CEO Tyler Denk. Articles discussed on this episode: https://www.theinformation.com/briefi... https://www.theinformation.com/newsle... https://www.theinformation.com/briefi... S…

    Read the verified transcript → Watch on YouTube ↗

  • Arena: $100M ARR in 8 Months

    I sat down with Anastasios Angelopoulos, CEO of ‪‪@ArenaAIOfficial‬ to understand how they became the neutral judge of the AI race. It started as a research project at UC Berkeley, with PhD students asking a simple question: are we measuring AI the right way? Today, Arena is the largest real-world AI evaluation platform in the world, reaching $100M ARR in under 8 months. In this interview, we got into: Why traditional AI benchmarks are fundamentally broken How Arena's leaderboard became the industry standard Agent Mode: millions of agentic traces, causal tracing How humans are actually usin…

    Read the verified transcript → Watch on YouTube ↗

  • Ο Αναστάσιος Αγγελόπουλος, CEO της Arena, στο «ΠρΑΙμ»

    Ο Αναστάσιος Αγγελόπουλος, CEO της Arena, καλεσμένος στο «ΠρΑΙμ» και τον Ηλία Σιακαντάρη. Official Sites: https://www.ert.gr/ https://www.ertnews.gr/ https://www.ertflix.gr/ https://www.ertecho.gr/ #ERTnews #ΕΡΤ

    Read the verified transcript → Watch on YouTube ↗

  • How to evaluate LLMs | the statistics behind Arena's rankings

    https://arena.ai Anastasios Angelopoulos, co-founder and CEO of Arena, presents a technical deep dive into how the platform evaluates large language models using live human preference data. Prior familiarity with probability and statistics is helpful but not required. The talk covers Arena's core methodology — pairwise comparisons, Bradley-Terry modeling, and Elo-based rankings — including the statistical foundations and mathematical derivations behind the ranking system. US Bureau of Labor Statistics’ Standard Occupational Classification (SOC) https://www.bls.gov/soc/2018/major_gr... Ope…

    Read the verified transcript → Watch on YouTube ↗

  • Arena Founder Anastasios Angelopoulos on AI Trends for 2026

    "Gone are the days where one number can describe how good a model is." We sat down with Anastasios, founder of Arena, to discuss what real-world usage data reveals about how developers are choosing models in 2026—and why the landscape is more diverse than ever. Key Insights: → 2026 will bring massive model diversity: sizes, domains, specializations → Every frontier lab will have multiple major releases this year → Different models excel in different categories: math, coding, creative, multi-turn → 5-8% of LM Arena users are "unhirable experts"—PhD-quality professionals providing real-wor…

    Read the verified transcript → Watch on YouTube ↗

Verified interview transcripts

Notable quotes

“When you put AI into the hands of humans and just when you do AB tests to try to see what's happening, you know, how it affects their actual workflows, you'll see that agents are lying to people. Agents are basically telling them what they want to hear. They're deleting their files in an unauthorized way. They're bypassing permissions and it's happening basically perniciously. All of us have experienced this even if ”

“Nobody should be looking at cost per token. Okay, why should you not be looking at cost per token, which is what the model providers release, by the way? They're releasing a rate card with cost per token on it. It's because, depending on the model, it's more verbose or less verbose, and so it's spending more tokens per task. So the cost per token is the cost that you're going to get per deliverable or per task is con”

“In the ideal world, Arena is the continual learning platform. The reason for that is that we just have a constant stream of fresh real world data that helps people improve for actual tasks as opposed to like some random benchmark that is made by oh like I created my private data set with like whatever you know whatever lawyers I could find to pay to like write me some legal documents or whatever. It's like that stuff”

“People are so fixated on cost per token. And at Arena, we really believe that you need to be evaluating things on cost per task. So even the way that we benchmark cost is something that we've continued to innovate on. Like here on the Pareto frontier for agent, you can see that it's the claws on the top. It's soul and then where's the Gemini model? I don't know. There it is over here. Not quite Pareto Frontier. Um, b”

“The fact that Kimmy was number one on our web development arena was huge national international news. But the fact that Kimmy is number one on web development, it's the first time really that we're seeing a violation of the distillation narrative. The distillation narrative being that the only reason why China is keeping up is because they're distilling American intelligence. But how can that be true if they're numbe”

“Anthropic revenue has been booming, booming, booming on the API, but to some extent, it's also easy come, easy go. It's like if you are spending that much money that quickly, there's no chance that you're not able to switch vendors. And so if an open source model comes by and it is able to do 90% of the work at 10% of the cost, which is totally feasible that that might happen over the next couple years and that certa”

The summary and quote tags on this profile are produced with AI assistance from verified, first-person interview transcripts, then checked by our team to confirm the speaker's identity and the accuracy of every quote. See how we verify →