oneshotlm
Compare all models on this prompt →

Flow-field particles · qwen3-vl-30b-a3b-instruct

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 · 4 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

I'll create a full-screen flow-field particle animation using p5.js. The animation will feature thousands of particles moving according to Perlin noise vectors, leaving fading trails with colors that shift over time.

write index.html (2970 bytes)
toolResult

Successfully wrote 2970 bytes to index.html

assistant

I've created a full-screen flow-field particle animation using p5.js. The animation features: 1. Thousands of particles that move according to Perlin noise vectors 2. Fading trails as particles move across the screen 3. Color shifting over time with a smooth gradient effect 4. Particles wrap around the edges of the screen 5. Responsive design that adjusts to window size The key components include: - A flow field generated using Perlin noise - Particle class with position, velocity, and lifespan attributes - Noise-based vector fields that guide particle movement - Color cycling through the hue spectrum - Fading trails created by reducing opacity over time The animation runs at full screen with no external dependencies, using only the provided p5.min.js file.

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

modelqwen3-vl-30b-a3b-instruct
vendorQwen
harnesspi-agent
generated2026-08-01

EVALUATION

3/5 working

Flow-field particle trails render and evolve over time, though very dim and low-contrast overall.

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 motion0.1%
on click0.0%
on drag0.2%
on wheel0.0%
on enter + space0.0%
on arrow keys0.1%
on w/a/s/d0.1%
frame spread2.4 / 255
console errors0
js errors none

METRICS

runtime13.0s
tokens in4.1k
tokens out1.2k
cached0
cost / run$0.0011