# AgentArena

> Benchmark your AI coding rig against the world.

AgentArena is a community-run, real-world benchmark of how agentic AI coding setups perform on shared tasks. Each test defines an ordered sequence of prompts; contributors run those prompts against their preferred agent · model · provider stack and submit per-stage scores. Aggregates roll up into a leaderboard, per-agent / per-provider / per-model breakdowns, and per-contributor profiles.

Snapshot: 6 tests · 10 runs · 30 stages · 1 contributors. Built 2026-06-11.

## Project

- [Overview](https://agentarena.tin.cat/): Project summary, top performers, latest activity.
- [Leaderboard](https://agentarena.tin.cat/leaderboard/): Agent · provider · model combinations ranked by average score.
- [Silicon beasts](https://agentarena.tin.cat/hardware/): Self-hosted rigs powering local inference — devices, GPUs, throughput.
- [Source & CONTRIBUTING.md](https://github.com/tin-cat/agent-arena): How to contribute a new test or a new run.
- [Raw data dump (JSON)](https://agentarena.tin.cat/stats.json): Complete machine-readable dump of every aggregation.
- [Full corpus (Markdown)](https://agentarena.tin.cat/llms-full.txt): Single-file expansion of every test prompt, contributed run, and catalog entry.

## Tests

- [Advanced event ticketing API](https://agentarena.tin.cat/tests/advanced-event-ticketing-api/): A backend-only event ticketing platform ("Frontstage") built with PHP and Symfony using Domain-Driven Design and a message-consumer archite…
- [Live message wall](https://agentarena.tin.cat/tests/live-message-wall/): A real-time, anonymous message wall web app with a TUI-inspired aesthetic. Visitors can post messages without logging in, and the wall upda…
- [Old-school demo](https://agentarena.tin.cat/tests/old-school-demo/): A 90s-style demoscene production built with Three.js: a single, self-running show of chained graphical effects synced to a MOD-like backgro…
- [Chat app with automated deployment](https://agentarena.tin.cat/tests/realtime-chat-with-deployment/): A WebSocket-based chat application with multiple rooms, presence, and basic moderation. Builds from a single shared room up to a moderated…
- [Snake game](https://agentarena.tin.cat/tests/snake-game/): The classic Snake arcade game implemented as a single-page web app. Later stages add progression, sound, and a local leaderboard, then mobi…
- [Titanic sinking simulation](https://agentarena.tin.cat/tests/titanic-sinking-simulation/): Physically-accurate simulation of the sinking of the Titanic. A quite complex, single-stage test (easy to run) that asks the model not only…

## Coding agents

- [Claude Code](https://agentarena.tin.cat/agents/claude-code/): Anthropic's official CLI for agentic coding in the terminal, with IDE integrations.

## Inference providers

- [Anthropic](https://agentarena.tin.cat/providers/anthropic/): First-party API for Claude models.

## Models

- [Claude Opus 4.7](https://agentarena.tin.cat/models/claude-opus-4.7/): by Anthropic. 7 run(s) across 6 test(s).
- [Claude Sonnet 4.6](https://agentarena.tin.cat/models/claude-sonnet-4.6/): by Anthropic. 1 run(s) across 1 test(s).
- [Claude Fable 5](https://agentarena.tin.cat/models/claude-fable-5/): by Anthropic. 1 run(s) across 1 test(s).
- [Claude Opus 4.8](https://agentarena.tin.cat/models/claude-opus-4.8/): by Anthropic. 1 run(s) across 1 test(s).

## Tech stacks

- [Node + Express](https://agentarena.tin.cat/stacks/node-express/): Node.js with the Express web framework. 4 run(s) across 2 test(s).
- [Three.js](https://agentarena.tin.cat/stacks/threejs/): Browser WebGL graphics built on the Three.js library. 1 run(s) across 1 test(s).
- [Vanilla JS](https://agentarena.tin.cat/stacks/vanilla-js/): Plain browser JavaScript with no UI framework or build step. 1 run(s) across 1 test(s).
- [PHP + Symfony](https://agentarena.tin.cat/stacks/php-symfony/): PHP on the Symfony full-stack framework. 1 run(s) across 1 test(s).

## Contributors

- [tin-cat](https://agentarena.tin.cat/contributors/tin-cat/): 10 run(s) across 6 test(s), avg score 0.83.