uni-robot-agent harness enables open-weight robot-use agents
Original title: Democratizing Robot-Use Agents: A Unified Harness with Open Models
Qwen3.8-Flash-Next reaches 31.39 RoboDojo score on open models, surpassing GPT-6 Astra-RoboProbe.
Robot-use Agents rely on proprietary models limiting research. uni-robot-agent harness supports further training on open-weight models without privileged inputs. Qwen3.8-Flash-Next performs competitively on 19 real-world tasks.
- Qwen3.8-Flash-Next scores 31.39 on RoboDojo
- Surpasses GPT-6 Astra-RoboProbe result of 28.97
- Competitive on 19 real-world tasks
- Effective robot use links to four general capabilities
- Supports training with open models
- Products
- uni-robot-agent
- AI models
- Qwen3.8-Flash-Next
From the original article
1 min read at huggingface.co
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.
By Jiarui Yuan, Yuxin Zuo, Xiaoyu Huang · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.