oneshotlm
Compare all models on this prompt →

Maze generator & solver · qwen3-next-80b-a3b-instruct

Generate a random maze with the recursive-backtracker algorithm on a canvas, then animate an A-star or BFS solver finding the path from start to finish.

⛨ sandboxed Open in new tab ↗
transcript · 4 messages
user

Generate a random maze with the recursive-backtracker algorithm on a canvas, then animate an A-star or BFS solver finding the path from start to finish.

assistant

I'll create a single-page web app that generates a random maze using the recursive backtracker algorithm and animates an A* solver finding the path from start to finish. Let me build this as a self-contained HTML file with CSS and JavaScript inline

write index.html (14123 bytes)
toolResult

Successfully wrote 14123 bytes to index.html

assistant

I've created a complete single-page web app that generates a random maze using the recursive backtracker algorithm and animates an A* solver finding the path from start to finish. The app is contained in a single self-contained `index.html` file with all CSS and JavaScript inline, and no external resources are used. The app features: - A maze generator using the recursive backtracker algorithm - An A* pathfinding solver with animation - Controls to generate new mazes, solve mazes, and reset - Visual feedback showing visited cells (light blue) and the solution path (green) - Start (green) and end (red) points clearly marked The maze is rendered on a canvas with a grid of 20x20 pixel cells, creating a 25x25 cell maze (500x500 pixels total). All code runs completely sandboxed with no network requests.

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-next-80b-a3b-instruct
vendorQwen
harnesspi-agent
generated2026-08-01

EVALUATION

3/5 partial

Maze renders with start/end markers, but no visible solver path (light blue/green) is shown despite Enter+Space triggering pixel changes.

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 motion0.0%
on click0.0%
on drag0.0%
on wheel0.0%
on enter + space30.3%
on arrow keys0.0%
on w/a/s/d0.0%
frame spread72 / 255
console errors0
js errors none

METRICS

runtime10.0s
tokens in6.4k
tokens out3.6k
cached0
cost / run$0.0046