GLM-5.3 in Claude Code 2.1.280
skills off
This is the unfinished output at the moment the run ended (Harness + model mismatch). See the review below.
Wall time 19m 24sCost $0.69 estimated
Review
Polished product shot on a dark stage: notch, a violet-to-coral wallpaper with a menu bar and dock, a strong warm screen glow with a reflection on the floor, a black keyboard with legible keys, a large trackpad, side ports and a silver unibody with a soft highlight. The lid sits slightly rotated relative to the base, so the hinge line and the two vanishing directions do not quite agree.
Verdict by the AI judge (provisional).
The full review
Harness + model mismatch. Reclassified 2026-09-24 (published as ok): the run ended on 'API Error: 400 This model does not support image inputs' when GLM-5.3 read its own render (…/shot1.png) with Claude Code's Read tool; the artifact was already written but the self-review was cut short (Claude Code x text-only model). Re-run on the text-only relay: macbook_pro_16__claude-code__glm-5.3__attempt2.
Attempt 1 of 2
Every try of this model in this harness on this prompt, oldest first. A later attempt is the cell's result: attempt 2.
- Attempt 1 Harness + model mismatch this run19m 24s$0.692026-09-24
- Attempt 2 OK the cell's result22m 06s$1.112026-09-25
Bugs
- lid rotated relative to base; hinge perspective off
Scores
Not scored: only runs that finished on their own are scored.
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
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 20 steps
Loading the timeline…
Download the full transcript Claude Code session JSONL, 479 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 $0.69, estimated from the token counts at list price.
| Tokens by type | Tokens | $ per 1M | Cost |
|---|---|---|---|
| Input (uncached) | 113,854 | $1.40 | $0.1594 |
| Cache write | 0 | $0.00 | $0 |
| Cache read | 738,497 | $0.26 | $0.1920 |
| Output | 77,190 | $4.40 | $0.3396 |
| of which reasoning | not recorded | billed as output |
- Total computed from tokens
- $0.6910
- Total reported by the harness
- none
Tokens summed from the Claude Code session JSONL (per message id, incl. subagent files); if Claude Code logged zero usage for this unrecognised model, the /cost screen's per-model usage is used instead. Claude Code tags glm-5p3 as unrecognized_model and its own $ figure has costBasis unknown, so it is NOT used: cost = tokens x Fireworks GLM-5.3 price.
Prices: Fireworks GLM-5.3 serverless price (STATE.md 22:42Z; same as the OpenCode/models.dev catalog entry fireworks-ai/glm-5p3), 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.
- Mode: interactive TUI in tmux (160x48), inside bubblewrap sandbox
- Isolation: bwrap: private /tmp, HOME and TMPDIR inside run dir, own CLAUDE_CONFIG_DIR copy; the project files, other runs, /tmp/hx_* not mounted