3D Pelican Bike Test: Opus 5.5 Beats GPT-6 Sol in Blender, $48 vs $35
Fun-Meaning-6474 · reddit · 2026-09-24
A Reddit user upgraded the classic "pelican riding a bicycle" SVG test to full 3D in Blender, giving Opus 5.5 and GPT-6 Sol the identical long art brief (Australian pelican, teal touring bike, sunset boardwalk, mackerel in beak) and having them build everything from an empty scene via Python scripts with no downloaded assets.
- GPT-6 Sol finished faster ($34.55) but fell apart on close inspection: broken leg IK rig (knees bending outward, feathers rotating with legs) and an empty, lifeless environment — failing to live up to stunning "leaked" footage.
- Opus 5.5 ($48.11) nailed the atmosphere, color scheme, and splash effects, with only a minor leg-base clipping issue.
Verdict: frontier models can now pass the pelican test in 3D, but the quality gap remains large. Setup: Claude API / Codex subscription + BlenderMCP.
More from coding & agent
- Why AI-made dev tools beat one-shot game generation: freedom of process — eschadiol · 2026-09-24
- RealSense VP on AgenticROS: Letting AI Agents Directly Control Physical Robots — chrismatthieu · 2026-09-24
- AWS API Gateway's Hard 10MB Upload Limit and the Presigned URL Fix — _jaydeepkarale · 2026-09-24
- Open-source Pragma gives coding agents a terminal-first workspace with Git worktrees — tech_w0rld · 2026-09-24
- Dev builds dense task annotation system with GPT-6 Astra, ships it as an LLM skill — chris_j_paxton · 2026-09-24
- Jev-as-a-Judge: hybrid agent eval flow escalates low-confidence calls to frontier models — omarsar0 · 2026-09-24