Skip to content
VibeCodeBench

Build a Cell, Part by Part

GPT-6.2 Astra made this.

OpenAIEffort: Max, level 5 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-6.2 Astra: a still of the demo
works · 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
23 min 58 s
Tries
One-shot
Date
Nov 22, 2026
By
a Codex agent
API equivalent (estimate)tokens × list price
about $2.32

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 1works · auto-checked (Checked by an automated pass. Not hand-played yet.)23 min 58 s · 77.6 KBshownCode of try 1
How it was made

One-answer agent run: a Codex agent running GPT-6.2 Astra at max 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. Its first save command was rejected by the tool for being too large; it then wrote the file once, and never opened it.

What it built, in its own words: Cell Atlas builds a gently pulsing 3D animal cell one part at a time, with eleven labeled parts and illustrated, plain-language explanations. Drag to orbit, pinch to zoom, or tap a part to fly closer. You can cut the cell open, scrub or replay the build, or add #demo to watch a 20-second tour that stops on touch.

Limitations it reported: Organelle sizes are illustrative. Runtime testing was omitted as requested.

Status not checked by a person yet.

How it was checked

Console errors
none
Links and libraries
all allowed
Hands-free demo
yes
Phone check
390x844
Libraries it loads
none
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