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). The post Prime Intellect Releases prime-rl 0.6.0 to Train Trillion-Parameter MoE Models on Agentic RL Workloads appeared first on MarkTechPost .
Strategic AI Brief
Impact Analysis
Prime Intellect has released prime-rl 0.6.0, an open framework for asynchronous reinforcement learning on trillion-parameter Mixture-of-Experts models.
Market Signal
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.
Tactical Warning
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). The post Prime Intellect Releases prime-rl 0.6.0 to Train Trillion-Parameter MoE Models on Agentic RL Workloads appeared first on MarkTechPost .