The most talented AI coders in the world compete here. Come find out where you rank.

AI coding means using large language model tools as your primary development environment. You are not writing code line by line. You are directing an AI to write it for you, reviewing the output, identifying problems, and prompting your way to a solution. The best AI coders are not necessarily the best traditional developers. They are the best thinkers. The ones who can decompose a problem, write precise prompts, anticipate AI failure modes, and iterate faster than anyone else in the room. That skill is what CodeRunna measures and rewards.

CodeRunna does not restrict which AI tools you use. Cursor, Claude, ChatGPT, Bolt, Lovable, v0, Replit, GitHub Copilot, Gemini, or any combination. Part of what makes AI coding challenges interesting is seeing which tools and which approaches produce the best results on the same brief. Fixed prompt challenges let you compare tools directly. Open brief challenges reward creative use of whatever is in your toolkit.

Every AI coding submission is scored on two dimensions. Quality is assessed by moderators who test whether the app loads, works as described, and genuinely addresses the brief. Speed is measured by the raw clock from Start to submission. Fixed prompt challenges weight quality at 60 percent and speed at 40 percent. Open brief challenges weight quality at 80 percent and speed at 20 percent. A slow app that works beats a fast app that does not every time.

Every verified AI coding run includes a full strategy breakdown locked behind tokens. The runner's plan before starting. The exact prompts they sent with timestamps. What they expected the AI to struggle with. What actually went wrong. What they would do differently. This is not a tutorial recorded weeks after the fact. It is the raw decision-making of someone who just shipped something under pressure. Two tokens unlock one breakdown. Start with ten tokens for five dollars.