oneshotlm
Compare all models on this prompt →

Flow-field particles · seed-2-1-turbo

Using the pre-provided p5.min.js (global p5, instance or global mode), create a full-screen flow-field particle animation: thousands of particles advected by Perlin-noise vectors, leaving fading trails, with colors shifting over time.

⛨ sandboxed Open in new tab ↗
transcript · 6 messages
user

Using the pre-provided p5.min.js (global p5, instance or global mode), create a full-screen flow-field particle animation: thousands of particles advected by Perlin-noise vectors, leaving fading trails, with colors shifting over time.

assistant
thinking
What I will create is a full-screen flow-field particle animation with p5.js.Check p5.js presence first.
bash {"command":"ls /tmp/ws/p5.min.js"}
toolResult

/tmp/ws/p5.min.js

assistant
thinking
Creating a flow field particle animation using global-mode p5.js is what I will do, with the effects of Perlin noise advection, fading trails, color shifting and full screen.
write /tmp/ws/index.html (4128 bytes)
toolResult

Successfully wrote 4128 bytes to /tmp/ws/index.html

assistant
thinking
It will be my next step to verify the file.

Created `/tmp/ws/index.html` — a full-screen flow field animation with: - **3000 particles** advected by a 3D Perlin-noise vector field (noise evolves in z over time) - **Fading trails** via translucent black rectangle overlay each frame - **Color shifting** — HSL hue drifts continuously, with saturation/brightness modulated by local particle speed - **Mouse interaction** — drag to repel particles, click to reseed ~600 particles - **Edge wrapping** + lifetime recycling so the field never thins out - Full p5 global mode, loading `p5.min.js` from the local file, no external resources

TAKE IT TO CODE

A single self-contained HTML file — copy it and open in any browser, or reuse the prompt with your own pi-agent run.

RUN

modelseed-2-1-turbo
vendorByteDance · Seed
harnesspi-agent
generated2026-08-13

EVALUATION

4/5 working

Dense flow-field particle trails animate continuously with color shifts, and inputs perturb colors/flow visibly.

Vision grade · anthropic/claude-sonnet-5

MEASUREMENTS

Raw readings from the capture, not a verdict — the grade above is the only judgement. Percentages are how much of the screen changed; the measure under-reports sparse motion on a dark background.

idle motion16.5%
on click36.1%
on drag22.7%
on wheel31.2%
on enter + space41.4%
on arrow keys23.0%
on w/a/s/d26.1%
frame spread21.9 / 255
console errors0
js errors none

METRICS

runtime33.3s
tokens in7.2k
tokens out1.9k
cached4.5k
cost / run$0.0060