DeepSeek V4.1 Flash in OpenCode 1.18.32
skills off
Zoom, inline player and more ways to view
Showing a WebP render of the 201 KB SVG so the page loads fast; zooming loads the SVG itself.
Wall time 10m 29sCost $0.092 reported6.5/10 · Draft score — AI judge
Review
Re-run after a launcher fault. A clean dark-stage render with a vivid pink-blue gradient wallpaper, the notch, a thin menu bar, speaker grilles, side ports and a soft purple light spill.
The full review
Re-run after a launcher fault. A clean dark-stage render with a vivid pink-blue gradient wallpaper, the notch, a thin menu bar, speaker grilles, side ports and a soft purple light spill. It has more polish than the same model in dsh, but less content: no dock, and the keyboard is a dark blurred block rather than legible keys. The lid reads near-frontal while the base is in three-quarter view, so the perspective doesn't quite agree.
Verdict by the AI judge (provisional).
OK. Finished on its own after 10.5 min and produced macbook.svg.
Attempt 2 of 2
Every try of this model in this harness on this prompt, oldest first. This run is the cell's result.
- Attempt 1 ended in an infrastructure failure, not a model result30m 32s2026-09-23
- Attempt 2 OK this run10m 29s$0.0922026-09-23
Bugs
- Lid and base perspective disagree (lid near-frontal, base three-quarter)
- Keyboard is a mushy dark block; individual keys barely read
- No dock or UI content beyond the menu bar
Scores
AI judge
6.5/10
Draft score — AI judge
Visitor votes and reviewer grades are not available yet.
| Criterion | Marks |
|---|---|
| macbook | 1.5 |
| perspective | 1 |
| screen glow | 1.5 |
| aluminium | 1.5 |
| craft | 1 |
| Total (sum of the marks) | 6.5 / 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), 373 characters. Every result for this prompt
Generate an SVG that looks like a realistic 3D render of a MacBook Pro 16" (opened, slight three-quarter view, screen glow, aluminum body). Use gradients and perspective to fake the 3D. Save it as macbook.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
Versions saved during the run
8 intermediate outputs, in the order the model wrote them.
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 15.1 MB; it is published as a scrubbed, compressed download rather than shown on the page.
Show all 136 steps
Loading the timeline…
Download the full transcript OpenCode session export, 131 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 $0.092, as reported by OpenCode (its own bill for the run). The list-price estimate from the token counts is $0.092.
| Tokens by type | Tokens | $ per 1M | Cost |
|---|---|---|---|
| Input (uncached) | 94,415 | $0.15 | $0.0142 |
| Cache write | 0 | $0.15 | $0 |
| Cache read | 6,807,168 | $0.00 | $0.0204 |
| Output | 32,468 | $0.60 | $0.0195 |
| reasoning | 63,234 | $0.60 | $0.0379 |
- Total computed from tokens
- $0.0920
- Total reported by the harness
- $0.0920
- Difference
- computed 0.0541 vs harness 0.0920
Tokens from OpenCode per-message usage; harness total = sum of OpenCode per-message cost. OpenCode reports reasoning separately; both bill at the output rate (output line here = output only, see reasoning).
Prices: api-docs.deepseek.com/quick_start/pricing (read 2026-09-23); off-peak rates, peak (weekdays 01-04 and 06-10 UTC) is 2x, 2026-09-23.
Run notes
- Playcode MacBook SVG Benchmark prompt verbatim, except its last sentence ("Respond with ONLY the SVG markup, no explanation.") is swapped for the save-to-file harness suffix, as the prompt research recommends for harness runs.
- run started off-peak: off-peak rates
- Mode: interactive TUI in tmux (160x48), inside bubblewrap sandbox
- Isolation: bwrap: private /tmp and /run, private PID ns; only this run dir, harness binaries (ro), opencode config, /usr, /etc bound; HOME/TMPDIR/XDG data private; the project files and other runs not mounted