Qwen Open-Sources QwenGyre, Speeding Up Long-Horizon Agent RL Training by 1.85x

Qwen has released and open-sourced QwenGyre, an end-to-end online RL training framework for extreme long-horizon agent tasks. Its Elastic Scheduler dynamically allocates GPUs, achieving up to 1.85x faster training, with an accompanying arXiv paper.

2026-10-12 ~ 2026-10-12 · 3 related posts

• Qwen releases QwenGyre: elastic GPU scheduling speeds long-horizon agent RL training by up to 1.85x — Alibaba_Qwen · 2026-10-12
• QwenGyre: Elastic RL framework trains xlong-horizon agents with up to 1.85x speedup — aigclink · 2026-10-12

1 near-duplicate retellings: ResearchCrafty1804