1

Hao AI Lab's Open Robot Agents Rival GPT and Claude on Visual Input

marketbrief.now

Hao AI Lab released ORA (Open Robot-Use Agents), an open-weight framework using the DeepSeek V4.1-Flash model to enable zero-shot robot control from visual inputs without specialized vision-language-action models. The system allows open models to compete with proprietary agents like GPT and Claude, though DeepSeek V4.1 is noted as less capable in vision-language modeling than some competitors.

Sign in to suggest

Key sources

  1. SOURCE@haoailab“An open harness powered by @deepseek_ai V4.1-Flash, ORA operates robots from visual input—zero-shot, no VLA—and competes with frontier-model agents”x.com
  2. SUPPORT@teortaxestex“Economically, V4.1 is the best Chinese model, but not remotely the best VLM”x.com
  3. SOURCEhuggingnewshuggingnews.com
Markdown