Zhipu.AI Open-Sources GLM: Models 8x Faster Than DeepSeek-R1 and Global Expansion Ahead of Possible IPO

Chinese AI company Zhipu.AI has open-sourced its next-generation GLM models, including the GLM-Z1 inference model that is up to eight times faster than DeepSeek-R1, achieving 200 tokens per second on consumer GPUs. The company also launched an international platform, Z.ai, and is expanding globally, potentially setting the stage for an IPO.

SRPO: Kuaishou's New RL Method 10x More Efficient Than GRPO for LLM Training

Researchers from Kuaishou's Kwaipilot team have introduced SRPO, a two-stage reinforcement learning framework that achieves DeepSeek-R1-Zero-level performance in math and code domains with only one-tenth of the training steps. SRPO addresses cross-domain optimization conflicts, low sample efficiency, and premature saturation through staged training and history resampling.