LMArena is Growing to Support our Community Platform

LMArena started as a scrappy academic project from UC Berkeley: just a handful of PhD students and undergrads working day and night on a research prototype. Today, we have two announcements:

  1. We are starting a company to support LMArena! LMArena will stay neutral, open, and accessible to everyone. We will focus on improving our open community platform for testing and evaluating LLMs.

  2. We’re releasing a beta version of a new site, beta.lmarena.ai. We’re doing our best to lower the error rate, improve the UI layout, optimize mobile, etc. Thanks for all the feedback. Keep it coming!

Why start a company?

We hope that a company will allow us to continue building this valuable community platform, and give us the resources to improve it significantly over what it is today. Some of us have recently graduated from Berkeley and want to keep working on LMArena, and we are hoping to build a team to provide even better service to our users and the AI community at large.

LMArena will be staying true to its original mission. It will remain a neutral, open platform for testing and evaluating AI models. Our leaderboard will never be biased towards (or against) any provider, and will faithfully reflect our community’s preferences by design. It will be science-driven. Staying neutral and earning community trust will always be essential to the success of our business. But this isn’t just about being strategic; it’s our choice, and part of our personal motivations for starting this company. So we will never deviate from that north star.

In fact, starting a company will accelerate our mission. As a first step, we’re finally addressing the feedback we’ve received and rebuilding our core platform, lmarena.ai. This means: fixing all the bugs, speeding up the experience, and adding features, like login, chat history, and personal leaderboards. It also means supporting more open research, like prompt-to-leaderboard, building more evaluations like WebDev Arena, RepoChat Arena, and Search Arena and working to improve our security. Moving forward, we’re excited to grow in a way that supports the community that helped us build this platform.

We haven’t fully worked out the business model, but we know AI companies want access to neutral and reliable evaluation services to speed up model development and improve real-world performance. This applies both to first-party model providers, and also to other companies for whom AI is a core part of their business. Our viewpoint is simple: organic, real-world human preferences will help drive progress towards more reliable AI, which will be valuable for everyone.

We’ll continue to bring the best AI models to everyone, and improve them through transparent, community-led evaluations. Today, we’re excited to launch the new LMArena Beta as the next step forward. With your feedback, we’re making big changes to the platform to make AI more accessible and reliable for everyone.

Launching Today: The New LMArena Beta

Over a month ago, we launched the Alpha version of a new LMArena directly with you in our Discord. So many of you have jumped in, tested its limits, and sent us an incredible stream of feedback, suggestions, and bug reports. Thank you!!

We want you to know: we’ve read it all — and you’ve made a real impact on LMArena. Many of the changes you see today are a direct result of what you told us, and we’ve been shipping improvements week over week thanks to your insights.

Beta is a more stable experience shaped by community feedback. It tackles many of the themes we heard again and again: improved error rates, better layout clarity, optimized mobile support, more intuitive voting, and improvements to the conversation flow.

What You Shared With Us

Through surveys, feedback forms, Discord threads, and spontaneous essays direct to our inbox, you’ve helped us understand what matters most:

Speed and performance

Clarity and design

We’ve taken this to heart. The new LMArena isn’t just a fresh coat of paint. It’s a full rebuild, with your ideas guiding the direction. The goal is to make LMArena more accessible, more reliable, and ultimately more fun to use for everyone testing the world’s leading AI models.

Try the Beta!

The Beta version is live and open to everyone, but this is still very much a work in progress. Beta means smoother, not perfect. That means:

We’re not done, and that’s where you come in! This is your chance to shape what the final experience becomes. Whether you love it, hate it, or hit a bug, we want to hear from you. Your voice is what helps us get it right.

If you’re new to LMArena, join us! Stay tuned for more exciting updates, we’re just getting started! For media inquiries, contact: [email protected]