~ same prompt, different LLM — every model gets the same brief in Blender, you judge the render ~
OpenAI / GPT 6 Astra / thinking:max has entered the arena NEW!
Curl-noise particle flowCategory: animationCapture: Prompt: v1Difficulty: ▮▮▮░░
TX: curl-flow.md — composed prompt
── block: build-request-open@v2 (start, global) ──

Task

Complete the following Blender task:

<task>

An abstract flow-field piece: thousands of points advected through a curl-noise vector field, leaving the impression of smoke-like ribbons of motion that swirl, diverge, and rejoin without ever scattering into chaos. Built with Geometry Nodes — a simulation zone (or equivalent iterative setup) moving points along a curl/noise-derived velocity each frame; instanced geometry or point rendering makes them visible.

The choreography matters more than the particle count: the field should have visible structure — two or three large vortices the flow orbits — and the motion should stay inside frame for the full duration. Color the points by speed or by which vortex owns them, against a near-black background, so the render reads as elegant data-art rather than a particle test. No physics-system particles; the motion must live in the node tree.

── block: build-request-close@v2 (end, global) ──
</task>
RUNS ON THIS TASKview in gallery →
RUNPROMPTCONTESTANTSTATUSCOST
r1 · 2026-07-16v1OpenAI / GPT 5.6 Lunaok$0.05
Votes cast on this bench: 000482 · 157 runs · 25 contestants
Best viewed with Netscape Navigator 4.07 at 800×600 · real runs, live ratings