Reddit user compares LLMs for parametric furniture modeling, finds Fable 5 best

The user tested GPT 5.5 and other models on identical prompts, measuring build time, token usage, cost, and quality. Fable 5 produced the best model at relatively low cost.