Weekly AI paper roundup covers world models, embodied AI, and self-improving agents
_akhaliq · x · 2026-07-26
Hugging Papers highlighted the top AI papers from July 20–26, spanning agents, embodied AI, world models, RAG, and code/data pipelines.
Notable papers in the roundup include:
- ABot-World-0: infinite interactive world rollout on a single desktop GPU
- RynnBrain 1.1: more capable, generalizable embodied foundation model
- TimeLens2: multimodal video temporal grounding
- RAGU: a multi-step GraphRAG engine with a compact domain-adapted LLM
- RESOURCE2SKILL: distilling executable agent skills from multimodal resources
- DataFlow-Harness: grounded code-agent platform for editable LLM data pipelines
- AREX: recursively self-improving agent for deep research
- EvolvingWorld and DeepSearch-World: co-evolving role-play agents and self-distilled deep search in verifiable environments
- Generative World Renderer at the Speed of Play
More from Embodied
- A OnePlus 12 can run Z Image Turbo and Flux.2 locally, but a 320×320 image still takes minutes — sgcego · 2026-07-26
- Orca opens a shared stack for dexterous hand research and teleoperation — tensorqt · 2026-07-26
- Jensen Huang says Optimus could help create a multi-trillion-dollar humanoid robot industry — XFreeze · 2026-07-26
- Alibaba and CUHK’s RynnWorld-4D predicts a robot scene’s future from one image — jiqizhixin · 2026-07-26
- Vivix A1 accepts voice, text and images mid-interaction, demo says — Scobleizer · 2026-07-26
- Vivix A1 demo shows synchronized speech, gaze, facial motion and body movement — Scobleizer · 2026-07-26