The Drowned Court

Get the best builds by email. No account needed.

Explore

Toys & Experiments

An interactive 3D world benchmark comparing competing AI agent builds.

byTimothée

Is this your creation?

PlayableOpen source

Description

An AI research experiment where multiple agent configurations (including GPT-6 Astra and DeepSeek) build the same interactive 3D scene from a reference image. Users can explore live snapshots of each model's multi-pass build process while comparing detailed metrics on time, API cost, and architectural fidelity.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: A rigorous and transparent comparison of multi-pass 3D world building across different high-frontier AI agents.
Attributed to: Timothée@tim_lb
Built with: GPT-6 Astra GPT-5.6 Luna
Links: ohmyunicorn.com
View tweet
View GitHub repo
Tags: #3d #benchmark #agents #threejs

Discovered: X
First posted: X · Sep 10, 2026
Listed: Added 49d ago
Build: Strong
Models: Strong

Get the best builds by email. No account needed.

Explore

Toys & Experiments