Traveling Salesman RL Environment is a Prime Intellect residency project that hardens a classic 10‑city TSP into a reusable, open-source RL/eval benchmark for LLMs: tool‑free prompts, a lenient parser that scores validity instead of format, and published hub runs (Gemini, Grok, Claude, Qwen, Kimi). It lives on Prime Intellect’s Environment Hub (setrf/traveling-salesman) and GitHub, showing how custom RL environments and synthetic data can push model reasoning beyond the usual scaling tricks.