AWorld: Orchestrating the Training Recipe for Agentic AI
This addresses the problem of slow and impractical training for Agentic AI systems, particularly in complex benchmarks like GAIA, by providing a scalable solution, though it is incremental as it builds on existing reinforcement learning and agent-environment interaction methods.
The paper tackled the bottleneck of inefficient experience generation in Agentic AI training by introducing AWorld, an open-source system that accelerates experience collection by 14.6x, enabling a Qwen3-32B-based agent to achieve 32.23% pass@1 accuracy on the GAIA test set, surpassing GPT-4o and rivaling DeepSeek-V3.
The learning from practice paradigm is crucial for developing capable Agentic AI systems, yet it is severely hampered by inefficient experience generation, a bottleneck especially pronounced in complex benchmarks like GAIA. To address this, we introduce AWorld, an open-source system engineered for large-scale agent-environment interaction. By distributing tasks across a cluster, AWorld accelerates experience collection by 14.6x compared to standard single-node, sequential execution. This critical speedup makes extensive reinforcement learning practical and scalable. Leveraging this capability, we trained a Qwen3-32B-based agent that achieves pass@1 accuracy of 32.23% on the GAIA test set, which surpasses GPT-4o (27.91%) and rivals DeepSeek-V3 (31.89%). Our open-source system and the resulting agent provide a practical blueprint for a complete agentic AI training pipeline, from efficient interaction to demonstrable model improvement.