Fable and 10 other LLMs refactor a LangGraph god node, Fable's proposal ranks first

Comparing Fable and 10 other LLMs on refactoring a LangGraph god node

The author gave 11 LLMs a 1,500-line LangGraph god node to refactor. Fable-5's proposal scored highest in peer review, followed by GPT-5.5 and DeepSeek-4-pro. GPT-5.4 and Opus-4.7 ranked near the bottom. Each model produced full code and architecture docs, then other models cross-evaluated them. Raw data and the ranking matrix are public. Caveat: this is one refactoring task, not a general coding benchmark, but it reveals clear differences in engineering taste across models.

Why it matters: A hands-on 11-model refactoring shootout with full code and peer-review rankings — not armchair commentary. Fable-5 taking first place is inherently discussion-worthy. Capped at 78 because it's a single-task personal experiment, not a controlled benchmark, so it stays at the f...