Nvidia Introduces Open-Weight Models
Nvidia launched new open-weight models designed for always-on agents and edge computing.
The Nemotron 3.5 Super VL is a large, 120-billion-parameter model geared for agents used by the enterprise. While it uses fewer parameters than the Nemotron 3 Ultra model, Nvidia states it can deliver similar accuracy in instruction-following and tool-use tasks.
Local Models for AI PCs
At the other end of the spectrum, the company also launched two lightweight Bolt models. These are intended to run locally on new AI PCs running the Nvidia RTX Spark chip and workstations running the DGX Spark. Such hardware, aimed at consumers, programmers, and enterprise knowledge workers, is being released to the market through partners like Microsoft and Dell.
Any organization that wants to run agents on Nvidia infrastructure now has an open, optimized model for their tier without having to choose between capability and deployability, Nvidia’s vice president of Open Models and Software, Kari Briski, wrote in the blog post announcing these models.
Performance and Enterprise Adoption
Nvidia claims that the Super VL completes tasks faster than competing models when post-trained on specific agentic tasks. These tasks include document intelligence, financial accounting, scientific experimentation, medical insurance claims, and security attack resilience.
Nvidia has backed its claims with a list of heavy-hitter companies that it says are using Super VL. These include Skydance, EY, Ramp, Crowdstrike, Perplexity, and Salesforce.
These are not the first open-weight models that Nvidia has released to compete with proprietary frontier labs. The company maintains a family of Nemotron models that handle multimodal artificial intelligence, reasoning, coding, math, and enterprise automation.
Shifting Industry Dynamics
However, fast-growing proprietary AI labs are now looking to chips and AI system makers beyond Nvidia to power their needs, utilizing hardware like AWS’s Trainium and Cerebras, as well as building their own custom chips.
As a result, Nvidia is leaning into the trend of enterprise companies wanting to run their own open models in private data centers. Unlike closed frontier models, organizations can post-train Super VL on proprietary data, deploy it on their own infrastructure, and retain complete ownership.
Privacy and Local Consumer Assistants
With the announcement of the smaller, local PC models, Nvidia is also pushing forward the trend of consumer AI assistants. Privacy-conscious consumers may hesitate to invite cloud-based agents into their lives with access to sensitive family, health, and financial information.
By creating machines that run models and agents locally, Nvidia hopes to enable users to delegate everyday projects without sending workloads to the cloud.
Of the two Bolt models, the Nemotron 3.5 Bolt 74B-A9B 24GB is a 74-billion-parameter model designed for local agentic tasks, balancing speed with reasoning to handle multistep tasks, tool calling, and web browsing. The smaller E4B model is aimed specifically at conversational agents.
Super VL and Bolt 74B-A9B are available now, with Bolt E4B scheduled to launch later in the month.
