**[Gradient Launches Distributed Reinforcement Learning Framework Echo-2 and Plans to Introduce RLaaS Platform Logits]** Foresight News reports that the distributed AI lab Gradient has released the Echo-2 distributed reinforcement learning framework, aiming to break the efficiency barriers in AI research training. This framework achieves a decoupling of Learner and Actor at the architectural level, aiming to reduce the post-training costs of large models. According to official data, the framework can reduce the post-training cost of a 30B model from $4,500 to $425. Echo-2 utilizes compute-storage separation technology for asynchronous training (Async RL), enabling sampling computation to be offloaded to unstable GPU instances and heterogeneous GPUs based on Parallax. The framework incorporates technologies such as bounded staleness, instance fault-tolerant scheduling, and the self-developed Lattica communication protocol, enhancing training efficiency while maintaining model accuracy. Additionally, Gradient plans to launch the RLaaS (Reinforcement Learning as a Service) platform Logits, which is currently open for reservations to students and researchers.