Arena has crossed $100M in annualized revenue run rate, eight months after launching our evaluation product. With our recent release of Agent Mode, millions of users on Arena are doing real work with agents, from coding to document analysis, in long-running, multi-turn sessions with hundreds of tool calls. Arena now evaluates objective criteria like task completion rates, hallucination rates, and more, far beyond our original human preference voting model. This expansion has taken us from a student project at Berkeley to one of the fastest growing companies in history. Go Bears! 🐻 Our core thesis is simple: to align AI with human values, we must directly measure its impact on people in the real world. Today's milestone is proof that Arena’s platform is the de-facto standard for post-deployment evaluation of AI.

RT @ProfKuang: My first Independence Day as a US citizen. I came here 2005 an awkward non-English speaking teenager. I naturalized exactly…

arena-rank just hit 100 stars :)

I come from a family of Greek immigrants that came to this country for a better life. @america has given us extraordinary opportunity to improve our situation. My grandfather was a waiter. He raised a doctor, an executive, and a professor. My father came as a graduate student to contribute to space science and technology. My wife is an amazing scientist and professor at Stanford, another immigrant from Serbia. I’d never have met her without the opportunities this country gave us. Great people from all over come to America to enjoy the freedom and meritocracy that you can’t get anywhere else. We celebrate our diverse cultures while achieving in our careers at the highest level, and ultimately doing something meaningful with our lives. Today is a great reminder to protect the values that make our country special. I am so proud to be a Greek-American. Happy 250, @america!

Claude Fable 5 is back - it might be the most impressive model we've ever seen. @petergostev ran it through 60 of the hardest 3D generations: modern cities, art, world wonders, and more. The results are incredible. Fable is clearly in a league of its own. Watch youtube here: youtube.com/watch?v=rTc2_-1K…

Now that Fable 5 is back in the Arena, check out this montage of a few of the prompts we put it through. Find it in Battle Mode and Agent Mode, and start testing it for yourself. Official leaderboard scores coming soon. youtube.com/watch?v=euiewern…

Since people have requested it, I've updated the leaderboard to include GPT 5.5. It seems to be a slight improvement over GPT 5.4 and comparable to Opus 4.8!

.@Arena ranks #8 in the "startup Arena", with 34.7M monthly visits! We're building the platform for measuring and advancing real-world intelligence at scale. Come join us!

Some big updates to GameDevBench! We’ve tripled (!) the benchmark to 333 tasks and updated the paper accordingly. We also added GLM-5.2 and Opus 4.8 to the leaderboard. Opus now leads by a small margin, while GLM-5.2 is the strongest open-source model. This is despite not having any visual capabilities! Agentic game development remains far from solved, but it’s also exciting to see more agentic game dev work starting to emerge. I’m hoping we’ll see rapid progress here soon.

Let’s go @arena and @supabase !

no, you don’t understand, they took Fable and they lobotomized her, and now nothings the same. it’s like i don’t even know her anymore, what have i done, i miss her so much

So cool that we can measure this! "The new classifier also comes at the cost of flagging benign requests more often during routine coding and debugging tasks." translated: -27 point on code arena everything else, vision, documents, expert tasks it's still fantastic

Basically, the CIs overlap, it is not nerfed, so far looks like it’s the same model.

Fullstack code arena is yet another amazing development in @arena’s agentic capabilities. Now you can make almost any app in Arena. Check it out!

Check out @arena on this list :)

Interesting result

Super proud to back @ml_angelopoulos at inception ahead of this amazing milestone. Congrats to team @arena.

berkeley had professors who are cto of tech companies. in addition to the many cofounder/ctos like stonebraker and stoica, we also have chenming hu, tsmc cto kurt keutzer, synopsys cto the industry to academia boomerang will be awesome.

Fable 5 is back in the Arena! When it first debuted, Fable 5 ranked #1 in Agent Arena: our benchmark for real-world, long-horizon agentic performance. Agent Arena evaluates models on millions of real tasks submitted by a global community of users, with access to web search, filesystem, and terminal tools to complete complex workflows. Rather than raw win rates, the leaderboard uses causal tracing to measure how much each model's outcomes outperform the average model. @AnthropicAI's latest model is also live in Text, Vision, Document, and Code Arena: Frontend. Come give it a try in Battle Mode and Agent Mode and contribute to the leaderboards.

Check out first impressions of Claude Sonnet 5 in the Agent Arena on YouTube with @petergostev. Scores coming soon. youtube.com/watch?v=hh2WLKa7…

Felicis has been an amazing partner in building @arena