GPT-6 Astra in Codex CLI 0.156.1
skills off
Zoom, inline player and more ways to view
Wall time 3m 11sCost $0.74 estimated9.5/10 · Draft score — AI judge
Review
Tidy, editorial-style vignette: pelican on a teal city bike with fenders, chain and a bell, one wing on the handlebar, foot on the pedal, on a cream background with a pale sun disc.
The full review
Tidy, editorial-style vignette: pelican on a teal city bike with fenders, chain and a bell, one wing on the handlebar, foot on the pedal, on a cream background with a pale sun disc. Done in about 3 minutes with 7 tool calls, the cheapest and fastest strong result in the row.
Verdict by the AI judge (provisional).
OK. Finished on its own after 3.2 min and produced pelican.svg.
Bugs
- Minimal scene (vignette, no ground context beyond a shadow)
- Far leg is hidden, so the pedalling reads from one side only
Scores
AI judge
9.5/10
Draft score — AI judge
Visitor votes and reviewer grades are not available yet.
| Criterion | Marks |
|---|---|
| pelican | 2 |
| bicycle | 2 |
| riding | 2 |
| composition | 1.5 |
| craft | 2 |
| Total (sum of the marks) | 9.5 / 10 |
- 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. A draft score until the reviewer checks it. Rubric: Wave-1 SVG scoring rubric (AI Model Reviewer).
- 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
agentic harness (save the file; self-render allowed), 234 characters. Every result for this prompt
Generate an SVG of a pelican riding a bicycle. Save it as pelican.svg in the current directory. You may render and inspect your own output (headless chromium is available at /usr/bin/chromium) and iterate on your own before finishing.
How it was made
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 machine-format transcript is published as a scrubbed download.
Show all 14 steps
Loading the timeline…
Download the full transcript Codex CLI rollout JSONL, 119 KB. 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 $0.74, estimated from the token counts at list price.
| Tokens by type | Tokens | $ per 1M | Cost |
|---|---|---|---|
| Input (uncached) | 21,403 | $10.00 | $0.2140 |
| Cache read | 108,430 | $1.00 | $0.1084 |
| Output | 8,343 | $50.00 | $0.4172 |
| of which reasoning | 2,780 | billed as output |
- Total computed from tokens
- $0.7396
- Total reported by the harness
- none
Tokens from the Codex rollout token_count events (Codex prints no dollar figure); reasoning is part of output.
Prices: Project price table; OpenAI Standard tier, requests whose prompt is >272K tokens bill at 2x input/cache and 1.5x output (applied per request where per-request usage is logged), 2026-09-22.
Run notes
- a well-known benchmark wording "Generate an SVG of a pelican riding a bicycle" (a full stop added before the harness suffix).
- per-request tiering: requests with a prompt above 272K tokens billed at 2x input/cache, 1.5x output
- Mode: interactive TUI in tmux (160x48), inside bubblewrap sandbox
- Isolation: bwrap: private /tmp, HOME and TMPDIR inside run dir, own CLAUDE_CONFIG_DIR/CODEX_HOME copy; the project files, other runs, /tmp/hx_* not mounted