Skip to content
VibeCodeBench

Build a Cell, Part by Part

GPT-6.1 Terra made this.

OpenAIEffort: High, level 3 of 5

Made by a Codex agent. It wrote the file in one go and never ran or fixed it.

Build a Cell, Part by Part by GPT-6.1 Terra: a still of the demo
broken · auto-checked (Checked by an automated pass. Not hand-played yet.)
Click it to play here.Playing here. Done gives the page back.

Receipt

Made in
Codex
Paid with
ChatGPT subscription
Time
2 min 8 s
Tries
1 try · broken
Date
Nov 22, 2026
By
a Codex agent
API equivalent (estimate)tokens × list price
about $0.03

Estimate at list API prices. I used my regular paid plan, not the API.

Prompt and answer tokens at list price, thinking included; the coding tool's own instructions are not counted.

Try

  1. Try 1broken · auto-checked (Checked by an automated pass. Not hand-played yet.)2 min 8 s · 18.4 KBshownCode of try 1
How it was made

One-answer agent run: a Codex agent running GPT-6.1 Terra at high effort got only the prompt and wrote the file in a single write. It did not open or test it, and nobody edited it afterwards.

What it built, in its own words: Built a single-page interactive animal cell with a step-by-step build, organelle descriptions, cutaway switch, rebuild controls, and a demo loop. The page is designed for mouse and touch.

Limitations it reported: The 3D view may fail to load because OrbitControls uses a bare module import without an import map; the file was not tested, as requested.

Status not checked by a person yet.

  • Try 1: Automated check: The 3D cell never loads because three.js cannot be resolved; only the page text and controls appear, with a "3D view couldn't load" message.

How it was checked

Console errors
2
Links and libraries
all allowed
Hands-free demo
no
Phone check
390x844
Libraries it loads
cdn.jsdelivr.net/npm/[email protected]/build/three.module.js, cdn.jsdelivr.net/npm/[email protected]/examples/jsm/controls/OrbitControls.js
See the code

The benchmark

Build a Cell, Part by Part

Watch a living cell build itself one part at a time, then tap any part to learn what it does.

I keep the exact prompts private for now.

Every AI on this benchmark