🚀Introducing Code Arena: the next generation of live coding evals for frontier AI models. Built to test how models plan, scaffold, debug, and build real web apps step-by-step. Try Claude, GPT-5, GLM-4.6 and Gemini in Code Arena today!

Why Code Arena? AI coding has entered a new paradigm. Models don’t just output code—they build real applications. They use tools to read/edit files, and iterate like real engineers. Code Arena captures this workflow end-to-end, and you can see how models plan, act, and iterate in real time. Model generations are then compared head-to-head by Arena users on their functionality, quality, and design.

Moving forward, Code Arena will power the new WebDev leaderboard. Coming next: support for multi-file codebase, React apps, and more advanced tools, bringing evaluations closer to real-world development.

🧑‍💻 A community of developers vote in Code Arena, collectively shaping what “frontier AI coding” means. This isn’t static benchmarking. It’s evaluation in the wild, judged by real Arena users in the real world. Read more about it on our blog:

The next generation of AI coding evaluation is live! 🧪 Test models. 📊 Evaluate results. 🗳️ Vote on apps and websites. 👉 Check out Code Arena now:

@aryanvichare10 cooked hard, this is awesome!

Please add an automatic browser notification to let users know WHEN the models have finished responding. No one wants to wait 10 minutes for the LLM to finish. After that, the apps expire and can no longer be rated!!

Wow! Just wow!

Code Arena sounds super cool, lmarenaai! It's about time we see how these AI models actually build stuff, not just theory, right?

Let's goo @aryanvichare10

This is dope! Congrats guys!

This is huge. Finally, a real way to see how these models actually code step by step.

good, smooth, real lmarena

Is the website inaccessible now?

the gap from 1 to 6 is huuuuge

I love this idea!

AI's so bad ass I love it.

its Free?

Something went wrong with this response, please try again.

Code Arena's live coding evals for frontier AI models is revolutionary—step-by-step testing for Claude, GPT-5, and Gemini!

This redefines AI performance benchmarking.

You guys are doing real good 👏👏

This document proposes an architecture for stable and aligned AI, inspired by collective psychodynamics, by transposing concepts such as mimetic memory, emotional thresholds, and CBD (Convergence-Tilt-Dissipation) regimes to artificial systems. It critiques current AI architectures for their reliance on exogenous feedback, leading to instability, reward hacking, and out-of-distribution fragility, and suggests cumulative internal constraints for endogenous stability. The universal formula P(t) = A ⋅ ψ(S,R,V,M,D,C) × [O(t) ⋅ D(t)] models observable behavior, with variables such as M (cumulative memory), V (activation volume), and D (latent/active divergences) to limit dangerous states. CBD regimes emerge naturally without explicit coding, via phases of dispersion, incubation, switching, persistence, and dissipation, illustrated by figures and minimal Python simulations in the appendix. In conclusion, a safe AGI requires a restrictive internal structure (memory, thresholds, endogenous feedback) rather than external optimization, as a proof of concept for advanced agents.

This sounds amazing! Excited to see how these advanced models perform in real-time coding scenarios. Can't wait to try Code Arena! 🚀

非常 棒的一个 网站，可以知道自己中意哪个模型。建议，预览 可以显示图片，可以上传，删除文件夹与文件

are you kidding me?

is down?

Fix the damn website already

Code Arena testing how AI models actually build is important. Real evals for real coding. @awenetwork_ai x402 infrastructure lets these models pay for compute, data, and services while building. Infrastructure meets intelligence. That's how we scale.

Please fix it again, I want to continue the photo that is not appropriate, but instead the photo is stuck, I can't continue what is appropriate, it is stuck on that photo, please fix it, otherwise the website will be quiet

#Gaming

I'm using lmarena, and so help..so powerfull

The old one's or the new upcoming Gemini ?? Will try it..

dude there's a fucking captcha popup showing up every goddamn second on this shitty site, it's impossible to solve, the website is broken as hell!

I like it

Hello. I use your programs. It's been impossible to work lately. A captcha started popping up after each message. I take a few minutes to solve it and waste time. Well, turn it off already. It's impossible to work. You'd almost guess. Please do something about it!

This feels like a much more honest way to evaluate AI coding. Watching models plan, debug, and iterate like real engineers is way more meaningful than static benchmarks. I’ve seen similar conversations in @awenetwork_ai about evaluating agents in real workflows, not just scores.