# @RobertHaisfield — 2026-06-17

♥1702 ↻237 · https://x.com/RobertHaisfield/status/2067351401840414818

Are AI agents shape rotators? In this new benchmark, we let the models play campaign puzzles in Opus Magnum, a puzzle game by @zachtronics.

Ironically, Claude Opus 4.8 performed poorly, being beaten by GPT-5.5, Gemini 3.5 Flash, and GLM 5.2. Claude Fable 5 crushed them all. https://t.co/0TzcFp32B6

tags: author:roberthaisfield, kind:tweet, model:claude-opus-4-8, model:fable, model:gpt-5-5, on:claude-opus-4-8, on:gemini-3-pro, on:gpt-5-5, year:2026
cited on: _dossiers/claude-opus-4-8.md, _dossiers/gemini-3-pro.md, claude-opus-4-8, gemini-3-pro, gpt-5-5