AI Model Reviewer

GPT-6 Astra in OpenCode 1.18.32

skills off

← Open in the gallerySame prompt, all models

Zoom, inline player and more ways to view
Poster frame of GPT-6 Astra's page
Open the original page

The model's own page, run in a sandboxed frame only when you ask (a page that needs its own origin opens in a new tab on a separate demo site). Phones get a full-screen player capped at 1.5x pixel density (2x on larger screens) so it stays smooth; the original page runs uncapped. It starts in its auto-tour (T switches to play mode on a keyboard).

Recorded from the live demo (stepped clock). Preview video recorded by this site from the live demo. The moving picture above is that preview; the button runs the real page.

OKMomijigarimax effortNeutral harness

Wall time 25m 34sCost $4.77 reported

Snapshot 3 of 3 saved during GPT-6 Astra's run
  1. Snapshot 1 of 3 saved during GPT-6 Astra's run
  2. Snapshot 2 of 3 saved during GPT-6 Astra's run
  3. Snapshot 3 of 3 saved during GPT-6 Astra's run
Still snapshots saved during the run (the moving picture above was recorded from the live demo): 3 distinct frames.

Generation timelapse

Poster for generation timelapse

Generation timelapse (the agent's screen during the run, sped up). This is not the finished video. Download video (525 KB).

Review

No verdict yet.

The full review

OK. Finished on its own after 25.6 min.

Scores

AI judge

Not scored

No rubric for this prompt

Visitor votes and reviewer grades are not available yet.

AI judge
AI judge: the test runner (model claude-opus-5-5), scoring by visual review of the 1600 px render against the pre-registered rubric. No AI-judge score exists for this result: there is no rubric for this prompt, so no score is shown and none is made up.
Visitors
Public voting with a blind reveal is coming: you score the output first, then see the other scores.
Reviewer
The reviewer's own grade appears here once it is added.

Scores compare results for the same prompt only. AI-judge boards rank runs within one prompt, equal totals sharing a rank; scores from different prompts are never averaged.

Downloads

Files actually published for this run. Original output, site-made previews and post-run conversions are labelled separately.

The exact prompt

as run, 1,341 characters. Every result for this prompt

Build an interactive Three.js scene of a single red Japanese maple (momiji) standing on a small rocky island in a still, misty lake. Its leaves are deep autumn red and orange; some drift down onto the water, where they float and leave gentle ripples. Low mist hangs over the lake and soft mountain silhouettes fade into the haze behind. A stone lantern stands beside the tree with a warm flame inside. Let the viewer interact: dragging across the water stirs the floating leaves and sends ripples through them, clicking the lantern makes it swing so its light sways across the water, and a dial turns the time of day from dawn through day and dusk to moonlight, changing the sky, the colour of the mist, the lighting and the reflections, with the moon and its reflection at night. Keep the mood calm and painterly, with a light wind moving the branches. Everything is built in code from Three.js geometry, shaders and materials; no external models or textures. Build it as index.html (plus any local asset files you create) in the current directory. Three.js may load from a pinned CDN. You may serve it locally and open it in headless chromium (/usr/bin/chromium, software WebGL, no GPU) to screenshot and inspect your own output, and iterate on your own before finishing. Close any browser you open as soon as you've taken your screenshot.

How it was made

Wall time 25m 34sFirst artifact at 7m 37s6 saved versions

Versions saved during the run

6 intermediate outputs, in the order the model wrote them.

Version 6 of 6
  1. Version 1 at 7m 44s thumbnailv1 7:44
  2. Version 2 at 12m 49s thumbnailv2 12:49
  3. Version 3 at 14m 56s thumbnailv3 14:56
  4. Version 4 at 17m 04s thumbnailv4 17:04
  5. Version 5 at 19m 42s thumbnailv5 19:42
  6. Version 6 at 24m 08s thumbnailv6 24:08

Step by step

The prompt, every agent turn and tool call (tool name and a one-line summary; inputs and outputs are not shown), then the final message. The full transcript is 7.2 MB; it is published as a scrubbed, compressed download rather than shown on the page.

Show all 68 steps

Loading the timeline…

Download the full transcript OpenCode session export, 95 KB, gzip. Scrubbed: anything that looks like a key, token, e-mail address or phone number is replaced, and account details and the test machine's file paths are removed.

Cost breakdown

Headline cost $4.77, as reported by OpenCode (its own bill for the run). The list-price estimate from the token counts is $4.60.

Tokens by typeTokens$ per 1MCost
Input (uncached)114$10.00$0.0011
Cache write66,003$10.00$0.6600
Cache read1,588,892$1.00$1.5889
Output47,060$50.00$2.3530
of which reasoning15,824billed as output
Total computed from tokens
$4.6031
Total reported by the harness
$4.7681
Difference
Harness-reported $4.7681 vs list-price estimate $4.6031 (-3.5%); the harness figure is the headline.

OpenCode 1.18.32 reported its own total; token counts x list prices shown beside it. Cache writes priced at the input rate (the price table has no separate cache-write price). The harness reports reasoning tokens separately from output; here output includes them (31,236 visible + 15,824 reasoning), and both bill at the output rate.

Prices: Project price table (vendor list prices), 2026-09-22.

Run notes

one prompt, agentic harness (self-review allowed)prompt rewritten after the 2026-09-24 box losslabelled re-run (of threejs_momijigari__opencode__gpt-6-astra)

  • frozen prompt (rewritten): engine/prompts_rewritten/threejs_momijigari.txt: written by us from the recorded description (zero results)
  • plain OpenCode config copy (providers only; no ECC instructions/agents)
  • opencode export written to a file (not piped)
  • Mode: interactive TUI in tmux (160x48), inside bubblewrap sandbox
  • Isolation audit: outside paths seen