Prime Intellect Releases prime-rl 0.6.0 to Train Trillion-Parameter MoE Models on Agentic RL Workloads

Summary

<p>Prime Intellect has released prime-rl 0.6.0, an open framework for asynchronous reinforcement learning on trillion-parameter Mixture-of-Experts models. It trained GLM-5 on SWE tasks at up to 131k sequence length, with sub-5-minute step times and 256 rollouts, on 28 H200 nodes. This breakdown covers the inference and training optimizations behind those numbers — FP8 inference, Wide Expert Parallelism, prefill/decode disaggregation, router replay, and 3-D parallelism (FSDP, EP, CP).</p> <p>The post <a href="https://www.marktechpost.com/2026/06/23/prime-intellect-releases-prime-rl-0-6-0-to-train-trillion-parameter-moe-models-on-agentic-rl-workloads/">Prime Intellect Releases prime-rl 0.6.0 to Train Trillion-Parameter MoE Models on Agentic RL Workloads</a> appeared first on <a href="https://www.marktechpost.com">MarkTechPost</a>.</p>