BAA-chmark
BAA-chmark — sheep-herding arcade
A short 3D game about herding sheep. Different models received the same prompt; results range from fully playable games to outputs that cannot be launched.
- 45 results
- 44 demos
- Top-ranked Claude Opus 5
AI benchmarks and playable demos
Concrete examples of how different AI models handle the same assignments, with original outputs, practical evaluations, and demos you can try directly.
Model benchmarks
Each benchmark preserves the original model outputs, screenshots, practical evaluation, and playable demos.
BAA-chmark
A short 3D game about herding sheep. Different models received the same prompt; results range from fully playable games to outputs that cannot be launched.
GTA
A more demanding browser open world with a city, cars, pedestrians, police, and missions. The same assignment produced sharply different quality.
BUNGALOW
A complete furnished house created in Blender. Results are evaluated for layout, editability, visual quality, and operation in the 3D viewer.
Blog
Longer write-ups about software development, local models, and experiments tested on real projects.
The first prompt already produced a huge, lively city. Another 52 corrective prompts over one weekend refined it into a fun, playable game.
Two practical tests of DeepSeek, Qwen, and Laguna on real tasks from an established codebase. The same baseline and prompt within each test, with results verified in the application.
Projects
A new-build property search engine for Slovakia. It tracks more than 600 active residential development projects and updates them through automated data collection.
A toolkit for building terminal UI applications. Python and C++ versions for macOS are in progress; the goal is to make text interfaces for developer tools easier to build.
Author and contact
A software developer with experience across desktop, mobile, and web, mainly in C/C++, Python, and JavaScript.
FAQ
Marek Dubovsky's personal site for practical experiments, AI model benchmarks, playable demos, and selected software projects.
Yes. Playable and interactive results include direct links to demos and 3D viewers.
Original benchmark artifacts remain preserved. Any fixes, follow-up prompts, or website-owned wrappers are disclosed with the result.
Alongside AI benchmarks, the site presents selected desktop, mobile, and web work, including UIMD and flatscraper.com.