Qwen做了个新Agent模型Paradigm II，不用额外训练就在终端、编码、搜索和工具调用任务上全涨分，尤其没见过的任务也管用。
Qwen发布Paradigm II，一种基于世界建模的Agent基础模型。它通过单轮环境预测直接测试于多轮工具调用任务，无需Agent强化学习或任务特定调优。在7项基准上均取得提升，域内Terminal-Bench 2.0提升6.3%、SWE-Bench提升3.4%、WideSearch提升12.8%。域外基准Claw-Eval提升11.3%、QwenClawBench提升9.7%、BFCL v4提升9.0%。世界建模将'先预测后行动'内化为可迁移的推理模式。
🧠 Paradigm II — Agent Foundation Model: world modeling as agent capability. Single-turn, non-agent...
🧠 Paradigm II — Agent Foundation Model: world modeling as agent capability. Single-turn, non-agentic environment prediction → tested directly on multi-turn, tool-calling agent tasks. No agentic RL, no task-specific tuning. Gains across 7 benchmarks, including 3 entirely out-of-domain: - In-domain: Terminal-Bench 2.0 +6.3, SWE-Bench +3.4, WideSearch +12.8 - Out-of-domain: Claw-Eval +11.3, QwenClawBench +9.7, BFCL v4 +9.0 World modeling internalizes "predict before you act" as a transferable reasoning pattern. 💬 1 🔄 1 ❤️ 32 👀 6892 📊 4 ⚡