Industry change
NVIDIA expands Nemotron 3.5 open models for agents
- Published
- Source
- NVIDIA Blog
Summary
NVIDIA released Nemotron 3.5 Super VL, a 120-billion-parameter open model for data centers, plus Bolt models for local and edge agents. Per the company, Super VL targets up to 5x faster task completion for always-on agents, while Bolt runs local agents on RTX/DGX Spark hardware without the cloud. NVIDIA says companies including Atlassian, CrowdStrike, and Perplexity are post-training the model on their own data. The performance figures and adoption examples are all NVIDIA's own claims.
Why it matters for our work
Open agent models that run from servers to local hardware widen the option of running agents on your own infrastructure instead of relying solely on external APIs. Self-hosted models are becoming more practical for data-sensitive research and workflow automation.
Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.