Playground · Animation · Canvas · Standing test: we add each new model as it launches
Three AI models carve a swan from a block of ice with a chainsaw and chisels, in one HTML file. Watch every answer live and move the light.
Ice is hard to draw: it is clear, it bends the light, and a carved swan is a solid object with a long neck that has to hold together. The prompt asks for the real tools, a chainsaw first and then chisels.
Each answer has to make the piece from start to finish in one HTML file, hold on the result, and let the pointer move the light, an idea we took from @kaolti's test of AI models embroidering a logo. Claude Opus 5.5 and Muse Wonder each had one try over the API at their default settings, with room for 128,000 tokens. Muse Wonder Full is not on the API yet, so it answered the same prompt in Muse Code, Meta's agent app, where it could screenshot its own page and fix what was wrong before handing it over. That is a different test from a single try, closer to how people use Full today; it has no metered time or cost, and its note says how many rounds it took.
Carve a swan out of a clear block of ice, the way an ice sculptor would with a chainsaw and chisels. Make it a single self-contained HTML file (inline CSS and JavaScript, no external libraries, fonts or images) that fills the window. Show the piece being made from start to finish, then hold on the finished result. Moving the pointer moves the light, so the surface catches it the way the real material would. Click to start again.Show the whole prompt
Default settings: no thinking effort set, no temperature, and as much room as each model has (its own output limit). A card says so when an answer was made with other settings: “limit 128k” means it could use up to 128,000.
| Rank | Model | Checks | Time | Cost |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 Our pick | 4 of 4 | 9.5 min | $1.154 |
| 2 | Muse Wonder Full | 3.5 of 4 | – | – |
| 3 | Muse Wonder | 2 of 4 | 83 s3 tries | $0.204 |
Ranked by our pick, then checks passed, then cost. Time and cost count every try; each dot is one check, in the order listed below (filled: passed, half: partly, empty: failed). Click a model for its notes.
The most sculptural answer: a block on an ice base is roughed out with a chainsaw, refined with a chisel, given feathered wings, then polished with a torch. The finished swan is carved in the round, with a curved neck and stacked wing feathers, and glints that move with the pointer. 9.5 minutes, $1.15.
Made in Muse Code, not over the API: Full could screenshot its own work and fix it, and took 8 rounds. The whole routine is here: a clear block, the chainsaw throwing chips, chisels, feather detail and a torch polish with an orange glow. But the swan is a flat side profile with a long low body and a thin, kinked neck, more goose than swan, and most of its rounds went on fixing that neck. The ice glints as the pointer moves.
A chainsaw traces a line across a flat block, then a glowing translucent outline of a bird appears, with wing lines drawn inside it. It reads as a duck more than a swan, and as a drawing on glass rather than carved ice. The light shifts its glow. Our first two requests were refused before it wrote anything (by our account's credit limit at its listed 943,718 tokens of room, then over a thinking setting this endpoint does not take); the third, with 128,000 tokens, is the answer. 1.4 minutes, $0.20.
One answer per model, at each model's default settings unless the card says otherwise, through OpenRouter. If an answer ran out of room or stopped short, we asked again with more room, then with thinking off, and say so. This is a quick look, not a benchmark: run it again and every answer will differ. For measured results, see our benchmarks.
How we pick. Claude Opus 5.5 tried every answer (clicking, dragging and playing it in a browser), marked the checks and drafted the notes. Our pick is a judgement among the answers with the most checks passed, not a score: the checks show what works, the pick is which of those we would rather use. Opus is an Anthropic model, and Claude answers are our pick in 17 of 28 tests, so weigh Claude picks with that in mind. GPT-6.1 Sol, shown anonymised stills and our check marks, put Claude Opus 5.5 first for its in-the-round volume and refractive ice, then Muse Wonder Full, whose neck it found awkward and flat, then Muse Wonder. We agree.
What we checked, by trying each answer ourselves: Shows the carving: chainsaw roughing, then chisels, then a finish; The finished piece reads as a swan carved in the round; It looks like clear ice, and it glints as the pointer moves the light; Finishes, holds on the finished piece, and starts again on a click. A partial pass counts as half.
Claude Opus 5.5 is the only one that carves: its swan is a 3D sculpture on an ice base, roughed out with a chainsaw, chiselled and polished clear with a torch, and the glints move with the light. Muse Wonder Full goes through every stage of the craft, chips flying, but ends on a flat, goose-like profile after eight rounds of fixing it in Muse Code. Muse Wonder drew a glowing outline of a duck.
Our pick: Claude Opus 5.5
Our pick: Claude Opus 5.5
Our pick: Claude Opus 5.5