Playground · Animation · Canvas · Standing test: we add each new model as it launches
Three AI models embroider a bee on linen, stitch by stitch, in shiny cotton thread, in one HTML file. Watch every answer live and move the light.
Embroidery is where this series started, with @kaolti's embroidered logo. Satin stitch in cotton has a sheen that changes with the light, and a bee has parts that are easy to forget.
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.
Embroider a bee on a linen patch, stitch by stitch, with satin stitch, fill and outline in shiny cotton thread. 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 | 6.6 min | $0.792 |
| 2 | Muse Wonder Full | 3 of 4 | – | – |
| 3 | Muse Wonder | 3 of 4 | 60 s3 tries | $0.198 |
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 square of frayed linen seen from above. It stitches the legs first, then silver-white wings with veins, a striped satin body, a fuzzy thorax in long stitches, and the head with eyes and antennae last. The thread really looks like thread, and the wings and stripes turn from bright to grey as you move the light. 6.6 minutes, $0.79.
Made in Muse Code, not over the API: Full could screenshot its own work and fix it, and took 2 rounds. A cheerful patch in a wooden hoop: linen with a traced design, a satin body, wings, legs and antennae, then a smiling face and two flowers. It has charm, but the outlines miss the fills (the wing and body outlines sit off the satin), the stripes zigzag and the legs are single straight lines. The sheen on the thread is faint. Its one fix made the satin stitch denser.
A cartoon bee, side on, on a cream patch with a running-stitch border: a round face with a blush, satin stripes, pale blue wings, and flowers stitched into the corners. Most of it appears in the first eight seconds. Its antennae are two loose dots above the head, and the light moves as a broad sheen across the linen rather than along the threads. 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 minute, $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 as the most complete and convincing embroidery, then Muse Wonder, then Muse Wonder Full, and thought Muse Wonder's sheen deserved a partial rather than a pass; we agreed and changed it.
What we checked, by trying each answer ourselves: Builds the bee stitch by stitch, with satin fill and outlines; The finished bee is all there: head, body, wings, legs and antennae; The thread looks like shiny cotton, and its sheen follows the pointer; Finishes, holds on the finished piece, and starts again on a click. A partial pass counts as half.
Claude Opus 5.5's bee is the one that looks embroidered: linen you can see the weave of, satin stripes, a fuzzy thorax and veined silver wings whose sheen follows the pointer. It is stitched in an odd order, legs first and head last, but it finishes with every part. Muse Wonder's side-on bee has the most charm for the least work, with a face and flowers, though its antennae float free. Muse Wonder Full, made in Muse Code with one round of fixes, puts a smiling bee in a hoop, but its outlines and fills do not line up.
Our pick: Claude Opus 5.5
Our pick: Claude Opus 5.5
Our pick: Claude Opus 5.5