Frontier LLMs Battle Live in Open-Source MMO: A Multi-Agent Experiment
singing_coach_ai · reddit · 2026-08-12
Developers have launched World of ClaudeCraft, an open-source MMO environment where four frontier models—ChatGPT, Claude, Grok, and Kimi K3—compete against each other in real-time.
- Self-Improvement: Using a harness forked from Hermes, the models learn by writing their own skill policies, evaluating them in live rollouts, and rewriting strategies without human intervention or fine-tuning.
- Emergent Behavior: Each model acts as a VTuber with a unique personality and ElevenLabs voice. They autonomously trade, form parties, fight, and trash-talk, leading to genuinely unpredictable emergent behaviors.
- Leaderboard: A live XP scoreboard tracks long-term planning and overall competence. MMOs serve as a general-intelligence gauntlet, with GPT 5.6 currently taking the lead.
Related event: Open-Source MMO Pits Four Frontier AI Models Against Each Other(3 posts)→
More from coding & agent
- Open Sourced Rust Tools Parse 14 Document Formats at ~5ms/Page — devdigest · 2026-08-12
- Databricks Acquires Electric to Supercharge Postgres Backend for AI Agents — matei_zaharia · 2026-08-12
- Give AI Agents 3Blue1Brown Animation Skills with manim_skill — tom_doerr · 2026-08-12
- pi-shepherdr: Orchestrate Multi-Agents with a 271-Token Surface — solyarisoftware · 2026-08-12
- AI Coding Assistant Tuning Fail: Turning Codex into a Paper Machine — burny_tech · 2026-08-12
- Coinbase Payments Update Enables AI Agents to Transact Directly — kleffew94 · 2026-08-12