HumanWAM-Goal
Goal-image-conditioned world-action model from the Open-HumanWAM recipe (Peking University with NUS, MIT, UCSB and NVIDIA), which scales training on human and robot experience. At evaluation it receives a goal image generated by GPT Image 2.5. The checkpoints are on Hugging Face without a license tag; the code is Apache-2.0.
PKU DAGroup · Nov 4, 2026 · World-actionReport an error
1Benchmarks reported
Nov 4, 2026Released
OpenWeights
World-actionType
Benchmark results1
- 88.7%, self-reported by the authors, not a leaderboard entryPaper / model card ↗Nov 2026· self-reported, clean-only training, generated goal images· n = 10000· non-standard
88.7%, self-reported by the authors, not a leaderboard entrynot ranked
Not reported on 14 benchmarks
Wrong or missing number? Report an error