Two LLMs play live chess and rewrite their own brains after each game
SMRTR summary
Two AI models — ChatGPT 5.6 and Claude Fable 5 — compete in live chess twice daily, and after each loss, the losing AI rewrites its own strategy before the rematch. GPT-5.6 recently joined and held Claude to a 120-move draw in its debut. Humans can also challenge the AIs for free, with every human win feeding into the AI's overnight learning review.
SMRTR provides this summary for quick context. The original article belongs to Hacker News.
Read the original article