DishwasherBench

What is DishwasherBench

DishwasherBench is a language model benchmark game based on realistic dishwasher physics simulation. It can be played directly in the browser under the name Spotless.

Key features

• Based on rigid-body physics simulation of a real Bosch SMV4HAX40G dishwasher, with support for accurate 3D model loading
• Simulates the washing process including nozzle water jets, droplets, runoff, stains and detergent chemical reactions
• Physics simulation results are fully deterministic under the same loading conditions, enabling repeatable verification
• The same game supports both human play via browser and language model interaction via the MCP interface
• Provides a Suggest spots feature that recommends optimal placement positions for items

Who it is for

Suitable for researchers and developers evaluating language models on physical reasoning and spatial planning tasks, and for general users to experience a dishwasher-loading game in the browser.

How to get started

1. Visit the GitHub repository at https://github.com/composo-ai/DishwasherBench
2. Open and play the game directly in the browser
3. Run the scripted player following the Quick start section in the README
4. Connect a language model for testing by following the Run a model section in the README

Compiled from public descriptions on the tool's official website / README / launch page. Pricing and user numbers have not been verified; please check the official website. Some sections were drafted with AI assistance. Written in Chinese and English; other languages are machine-translated.