PREVIEW · the site is new and still changing
Nerdict
GPT-6 Lunamaxin Codex CLIMotion intro
Score16/17missed: Everything inside the 5% safe area from 4.2 s
Cost$0.009
Time2m20s
Runs of this setup#1 16/17#2 17/17#3 17/17
Motion intro by GPT-6 LunaOpen it running ↗

This is the exact file the model wrote, unedited, served from an isolated domain with the network blocked, as in the test.

What the robot did

The comparison clip (the same script for every model)

Rendered frame by frame (30 fps)

Checks

16/17

Written from the prompt (test v1, checks v4). The model never saw them.

  • ✓Opens without a page error
  • ✓Test hooks present (duration, seek, play, six beats)
  • ✓The line spans the frame at 1.25 slargura 1920 px aos 1,25 s
  • ✓Title types letter by letter on time (±20 ms)letra a letra no tempo certo
  • ✓Cards slide in from the right sides and settle side by sideaos 2,9 s visíveis: A=False B=False; aos 3,3 s A em 217 e B em 1703; aos 4,3 s A em 608 e B em 1312; lado a lado sem sobrepor: True
  • ✓Card texts and the COST PER TASK labeltodos presentes
  • ✓Counters count up from $0.00 to the final valuesobem de $0.00 e mostram $0.21 e $0.07 a partir de 8,0 s
  • ✓Title switches to DIFFERENT BILL. by 9.6 s8.9s 'SAME PRICE.' · 9.62s 'DIFFERENT BILL.' · 11.9s 'DIFFERENT BILL.'
  • ✓No text ever touches other textnenhuma
  • ✗Everything inside the 5% safe area from 4.2 s79: 4.2s title; 4.3s title; 4.4s title; 4.5s title
  • ✓Nothing shrinks: big title, wide cardstítulo 119 e 119 px de altura (mín. 86); cartões 640 e 640 px (mín. 422)
  • ✓Seeking to a time always gives the same framemesmo quadro
  • ✓Never stands stillsempre em movimento
  • ✓play() runs in real timetoca
  • ✓No uncaught exception during the session
  • ✓No console errors
  • ✓No network attempts
Model
GPT-6 Luna gpt-6-luna
Reasoning effort
max
Started
Oct 9, 19:30 UTC
Harness
Codex CLI 0.162.0, by OpenAI (not ours) · one shot · tools used: 0
Harness settings
isolated container with no global instructions, read-only sandbox; shell, browser, apps, multi-agent and web off; empty /work folder · imagem=nerdict-bancada-codex:0.162.0 · sandbox=read-only
Tokens
in 9,495 · cache read 0 · out 15,758 · reasoning 12,300 · requests 1
Cost
$0.009 API-equivalent · extra spend $0 (subscription)
Test
Motion intro v1 · checks v4 · prompt 6c307dbd2222
Output file
10.3 KB · sha256 9bdcdb917cc3
Run id
20261009T193027Z_vinheta-v1_luna-max_r1