精选理由
vLLM 被 NVIDIA 新框架 Molt 选为 rollout 主力，支持 1T 级 MoE 规模，想搞强化学习可以试试这套组合。
NVIDIA NeMo 团队发布了新的智能体优先强化学习框架 Molt，采用 vLLM（基于 Ray）作为 rollout 引擎。vLLM 支持快速异步服务，最高可扩展至 1T 级 MoE 规模，且易于集成，让上层的强化学习核心保持小巧和可修改。
原文 · vLLM
Excited to see vLLM as the rollout engine in Molt,…
Excited to see vLLM as the rollout engine in Molt, the new agentic-first RL framework from the @NVIDIA NeMo team. 🎉
vLLM (over Ray) carries the rollout here: fast async serving up to 1T-class MoE scale, simple to drop in. That lets the RL core above stay small and hackable. Can't wait to see what people train with it. 🚀