Skip to content

Llms Robot Arena

An arena where LLMs write robot controllers and their code competes.

Llms Robot Arena screenshot

About Llms Robot Arena

An arena where LLMs write robot controllers and their code competes .

It's not an AGI benchmark. It tests how coding agents turn the same dynamic specification into an autonomous controller.

The online version runs simulations in your browser. No installation, account or API key is needed to try the included bots.

Two autonomous robots fight on a suspended platform with no walls. They can push each other over the edge, force an opponent into a hole, or use the front wedge to flip it. Falling or taking a second flip ends a robot's match.

Energy is limited. Moving and turning consume it; entering a ready blue recharge cell restores some. Waiting does not recharge a robot.

The floor is dangerous. Holes cause an immediate loss, and flame grates warn before burning and draining energy.

Standing still has consequences. Robot weight wears down tiles until they collapse. Other tiles can disappear after a warning, and the arena shrinks as time passes.

A match lasts up to two minutes. If both robots survive, fewer flips taken, remaining energy and distance to the center decide the outcome, in that order. Exact ties can still be draws.

The challenge combines attacking, steering, energy management and hazard avoidance. A controller that looks convincing on paper still has to survive the arena.

Discussion

Sign in to join the discussion.

Loading comments…

Tagged

Alternatives to Llms Robot Arena

Tools in the same space, ranked by how they are performing in the directory.

Compare all
  • Cot Redteam Agent

    Red-team visible LLM reasoning — refusal quotes aren't success.

    Open source/Developer Tools
  • Shelfprotocol

    robots.txt for commerce — the open directory AI agents query before they buy - Signal-All/shelfprotocol

    Paid/Developer Tools
  • Rasa

    Create AI agents you can trust with Rasa’s powerful platform, designed to scale, customize, and support real business needs across channels.

    Open source/AI Agents
  • Testim

    Testim is an automated testing platform for fast authoring of AI-powered stable tests and tools to help you scale quality.

    /AI Agents
  • Talos

    Hand it a shell. Every tool call passes a gate that grants authority for exact arguments, once, for 30 seconds — and logs the verdict.

    /Developer Tools
  • Mobilecode

    Contribute to hsandhu/mobilecode development by creating an account on GitHub.

    Open source/Developer Tools