Playground · Animation · Canvas · Standing test: we add each new model as it launches
Three AI models throw a bowl on a potter's wheel and glaze it, in one HTML file, with light that follows the pointer. Watch every answer live.
A bowl thrown on a wheel is a process most people half know: centre the clay, open it, pull up the walls, trim, glaze. We wanted to see whether a model shows those steps or simply morphs a lump into a bowl, and whether the glaze catches the light the way glaze does.
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.
Throw a bowl on a potter's wheel from a lump of wet clay, then glaze it. 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 | 7.4 min | $0.900 |
| 2 | Muse Wonder Full | 3.5 of 4 | – | – |
| 3 | Muse Wonder | 2 of 4 | 2.7 min3 tries | $0.297 |
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.
A wheel seen from above at an angle, drawn in 3D. A cone of terracotta clay is centred, opened and pulled into a bowl with throwing rings, dried to leather-hard, then glazed in a celadon that pools in the rings, with a pale rim. Moving the pointer swings the highlight and the bowl's shadow round the wheel. 7.4 minutes, $0.90.
Made in Muse Code, not over the API: Full could screenshot its own work and fix it, and took 2 rounds. A side view with the inside showing: the clay is centred, coned up, opened, given a floor, pulled up and shaped, the foot trimmed, then the bowl is dipped in glaze and fired to a speckled celadon. The sheen and the shadow follow the pointer, but the shading is flat beside Opus's bowl. Its one fix removed a pair of hands it had drawn.
Two cartoon hands centre the clay and pull up the walls, then glaze pours in from a jug. But the piece ends as a teal funnel on a narrow foot rather than a bowl, with glaze drips drawn as straight bars down the outside and a glassy ring for a rim, and the light barely moves on it. 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. 2.7 minutes, $0.30.
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 of the three answers (mid-build, finished, and with the light moved) and our check marks, put Claude Opus 5.5 first for its depth, throwing rings and glossy glaze, Muse Wonder Full second and Muse Wonder third. We agree.
What we checked, by trying each answer ourselves: Shows the real steps: centring the clay, opening it, pulling up the walls, then glazing; The finished bowl looks like a real thrown and glazed bowl; Moving the pointer moves the light, and the glaze catches it; Finishes, holds on the finished piece, and starts again on a click. A partial pass counts as half.
Claude Opus 5.5 makes the bowl you would believe came off a wheel: a 3D form with throwing rings, a foot and a celadon glaze that pools in the rings, and the highlight swings round as you move the pointer. Muse Wonder Full shows the most complete process, from centring the clay to firing, and ends on a believable bowl, but it is drawn side on and flatter. Muse Wonder, in one try over the API, ends on a funnel with cartoon hands.
Full's answer was made differently: in Muse Code it could look at its page and fix it, which it did once here, so it is not a like-for-like comparison with the single tries.
Our pick: Claude Opus 5.5
Our pick: Claude Opus 5.5
Our pick: Claude Opus 5.5