Oriel — AI & technology worth noticing, every day.

Democratizing Robot-Use Agents: A Unified Harness with Open Models

Paper · https://arxiv.org/abs/2611.01560

Robot-use Agents (RUAs) have shown promising zero-shot manipulation capabilities, but reliance on frontier, proprietary models limits research into further training. To democratize RUA research, we introduce uni-robot-agent, a unified harness for competitive open-weight robot use that supports further training. Without privileged inputs, an out-of-the-box Qwen3.8-Flash-Next achieves a RoboDojo Score of 31.39, surpassing the published GPT-6 Astra-RoboProbe result of 28.97 and many VLA/WAM baselines, and performs competitively on 19 real-world tasks. Our analysis across model generations, families, and sizes links effective robot use to four general capabilities, suggesting that future RUA training should develop them jointly.

First seen by Oriel on .

AI summary

The paper introduces uni-robot-agent, a unified harness for training and evaluating open-weight robot-use agents, and reports strong results from Qwen3.8-Flash-Next.

  • Without privileged inputs, Qwen3.8-Flash-Next scores 31.39 on RoboDojo, exceeding the published GPT-6 Astra-RoboProbe score of 28.97.
  • The model also performs competitively across 19 real-world tasks.
  • The harness is designed to support further training and broaden research beyond reliance on proprietary frontier models.
  • Analysis across model generations, families, and sizes links effective robot use to four general capabilities, which the authors suggest should be trained jointly.

#robotics #robot-use agents #open-weight models #robot manipulation

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Points

Score reported by the source.

  • Hugging Face Papers: 49, latest observation ; +11 over observed period since

Comments

Comment count reported by the source.

  • Hugging Face Papers: 2, latest observation ; +1 over observed period since

Aliases

SourceExternal IDURL
Hugging Face Papers2611.01560https://arxiv.org/abs/2611.01560 Read discussion