According to @godofprompt, pairing Opus 4.7 for planning with GPT-5.5 for execution delivers superior 2026 coding workflow results.

According to Ethan Mollick on Twitter, GPT-5.5 in Codex generated a tabletop RPG Game Master’s Guide and Player Guide and self-reported playtesting of the ruleset, emphasizing narrative-driven mechanics and exhibiting some LLM-style patterns (source: Ethan Mollick). As reported by One Useful Thing, the project demonstrates end-to-end content generation, rules balancing, and iterative refinement within a single AI workflow, suggesting new opportunities for rapid prototyping of game systems and transmedia IP development (source: One Useful Thing). According to One Useful Thing, the approach lowers production costs for indie studios by automating lore bibles, encounter tables, and rule clarifications, and can be adapted to produce campaign supplements, scenario packs, and marketing copy, accelerating go-to-market cycles for TTRPG products (source: One Useful Thing). According to Ethan Mollick, the model’s strengths lie in storytelling cohesion and structured deliverables, while limitations include residual LLM artifacts that require human editing, indicating a viable human-in-the-loop model for commercial tabletop publishing (sources: Ethan Mollick, One Useful Thing).