AI coding models are evolving faster than ever before. The question is no longer simply whether a model can write static code, but how well it can build real end-to-end applications. That’s why we’re evolving Code Arena to incorporate fullstack capabilities.
When we first launched Code Arena last year, our goal was to change how AI coding models were evaluated. Seeing the positive response and adoption from our community, we’re now doubling down with the addition of measuring fullstack development abilities of the AI models. Starting today, Code Arena evolves from solely a frontend prototyping tool to a complete, fullstack AI development platform. The following is a look at what this new era of Code Arena unlocks for developers and builders.
What’s Different: The Fullstack Leap
We initially introduced Code Arena to measure "coding in motion," as developers could watch models build frontend web apps step-by-step, then vote on their performance & benchmark in the Code Leaderboard. With today's launch, we are continuing to innovate, evolving our architecture to support true fullstack web development, complete with databases, API keys, and live deployments.
With Fullstack, we now support building more complex apps, including:
- Apps that have a sign up or login flow, like a members-only ecommerce app where you need to login to browse and purchase
- The ability to connect to third party services via an API key, such as bringing in an OpenAI key to make API calls for an AI chatbot product.
- Apps that store a user’s progress for them to come back to, like an education or learning app where a user continually logs in over multiple days or weeks to pick up where they left off.
This changes Code Arena from more of an experimental prototyping environment to a daily-use tool where real work gets done.
Key new features and benefits include:
Database Integration: Database layer allowing agents to generate code for PostgreSQL, user authentication, and Row Level Security.Third party access: The ability to safely connect to third party services and apps, such as calling an LLM, or connecting to a payments API to accept payments for an ecommerce app, for example.Persistent Dev Server & Visual Terminal: A live dev server running inside the sandbox with hot re.Bash and Web Search Tools: We’ve moved beyond reading and editing files — Agents are now equipped to run any bash command, as well as search the web for real-time information or documentation via search, vastly expanding what’s possible to build.Fast Deployments: A seamless build-to-ship pipeline that allows you to deploy your fullstack web apps directly to Vercel.
**Empowering Builders To Do More **
By removing the limitations of the single-page frontend, Fullstack Code Arena is now a daily-use tool designed for three key audiences:
1. For Entrepreneurs and Small Businesses: Code Arena removes the technical barriers that limit your potential. By providing full data persistence and user management, we enable you to transform ideas into complex, functional applications without hitting technical ceilings.
2. For Developers: Code Arena provides a production-grade environment built for real work. With persistent, long-running sandboxes, you can seamlessly build, iterate, and ship complete applications within a single, integrated workspace.
3. For AI Model Labs: Code Arena delivers the high-fidelity evaluations necessary to advance software-engineering AI. By testing models on complex backend tasks—such as database operations and multi-file generation—you gain the critical data required to train the next generation of intelligent systems.
We’re excited to usher in this next stage of AI coding, which isn't just about just writing code—it's about building complete, real-world software.
Ready to build? Try out the new Fullstack Code Arena today.
Have additional questions? Visit our help center for more.