WildAWM open-sources 341 environment–task pairs; GPT-6 Astra scores ~72.7 pass@8
Fully open-source synthesis of 341 environment–task pairs where GPT-6 Astra reaches only ~72.7 pass@8; Qwen3.6-35B-A3B trained solely in these worlds gains +10.58 points across 8 OOD benchmarks.
🔥 Excited to share Agent World Model in the Wild (WildAWM)! 🌍 Fully open-source synthesis 🔓: No proprietary models needed! We synthesized 341 high-quality environment–task pairs, addressing 3 challenges in task synthesis: ✅ Solvability: At least one agent can successfully complete each task. 🔥 Difficulty: Challenging even for frontier models, with GPT-6-Astra achieving only ~72.7 pass@8. 🎯 Verifier reliability: 95% agreement between task verifiers and agent-as-a-judge ensemble. 📈 Effective RL training: Trained exclusively in these synthetic worlds, Qwen3.6-35B-A3B achieves an average +10.58-point improvement across 8 OOD benchmarks!