Nvidia has released a powerful open-source artificial intelligence model that competes with proprietary systems from industry leaders like OpenAI and Google.

The company’s new NVLM 1.0 family of large multimodal language models, led by the 72 billion parameter NVLM-D-72B, demonstrates exceptional performance across vision and language tasks while also enhancing text-only capabilities.

“We introduce NVLM 1.0, a family of frontier-class multimodal large language models that achieve state-of-the-art results on vision-language tasks, rivaling the leading proprietary models (e.g., GPT-4o) and open-access models,” the researchers explain in their paper.

Research videos aren’t known for their narrative arcs. Occasionally, however, you see one that keeps getting better the longer you watch. “Beyond Manual Dexterity” is one such video. It wows you in the first few seconds with a highly dexterous robotic gripper, and then things start to go off the rails.

The video debuted along with a research paper of the same name at IEEE’s International Conference on Robotics and Automation in Rotterdam this week. The work, conducted by Switzerland’s EPFL, explores ways in which robotic hands can grow beyond existing confines to grasp an even wider range of objects.

“The deep learning model has significantly advanced dexterous manipulation techniques for multi-fingered hand grasping,” the team notes. “However, the contact information-guided grasping in cluttered environments remains largely underexplored.”