Qwen introduces QwenGyre for elastic RL in extreme-long-horizon agent training

What happened

Qwen introduces QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. Its Elastic Scheduler reallocates GPUs between rollout and training without interrupting live executions, while its Trajectory Processor converts branching executions into prefix-shared trajectory trees with bounded, deduplicated training samples. Scaled to Qwen3.8-2.4T-A95B with ~700K tokens per rollout, QwenGyre raises NL2RepoBench from 52.5% to 58.5% in 48 steps and cuts end-to-end training time by ~44% vs Async. Paper: https://arxiv.org/abs/2609.33848

The story so far

Qwen introduced QwenGyre, a reinforcement learning system for extreme-long-horizon agent training, and reported that it raised Qwen3.8-2.4T-A95B’s NL2RepoBench score from 52.5% to 58.5% in 48 steps while cutting end-to-end training time by approximately 44% compared with Async. The reported training setup used approximately 700K tokens per rollout. QwenGyre’s Elastic Scheduler reallocates GPUs between rollout generation and training without interrupting live executions. Its Trajectory Processor converts branching executions into trajectory trees that share common prefixes, producing bounded, deduplicated training samples.

Keep reading

Trending now

1. 1Meta Releases Muse Spark Open With Mac and Cloud Support, Topping US Open-Weight Index
2. 2Google Launches Gemini 4 Argon Globally, Expands Paid Access as Benchmark Results Vary
3. 3OpenAI Updates GPT-6 Astra Pro in ChatGPT With Parallel Agents and Faster Responses
4. 4Meta’s New Business Terms Raise Dispute Over Website Crawling and AI Training Rights