Nvidia releases Nemotron 3.5 and Bolt open-weight AI models

By Splitfeed EditorialLast updated Oct 20, 2026, 01:20 UTC

On October 19, 2026, Nvidia released three open-weight models: Nemotron 3.5 Super VL and two models under the Bolt name. Nemotron 3.5 Super VL is aimed at enterprise multimodal workloads run inside company-owned data centers, while the two Bolt models are designed specifically for local AI agent inference running directly on personal computers. The release places Nvidia on both ends of the on-premises spectrum, from large corporate infrastructure down to individual user hardware.

The broader significance of this release sits in the ongoing shift away from cloud-dependent AI deployments. Enterprises have grown increasingly cautious about sending sensitive workloads to third-party cloud providers, and open-weight models that can be self-hosted give those organizations more control over data and costs. For the consumer side, on-device AI agents represent a different kind of privacy proposition, where inference happens locally and no data leaves the machine. Nvidia, which already dominates the hardware layer of AI infrastructure, is now pressing further into the software and model layer, which raises genuine questions about how much of the AI stack a single company should control and whether open-weight truly means open in practice, given that the weights are released under terms Nvidia sets.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, have published reactions to this release yet. Typically, the Pro-AI camp would welcome the move as democratizing access to capable models and enabling privacy-preserving AI at both enterprise and personal scale. The Anti-AI camp would likely raise concerns about accelerating AI adoption without adequate safeguards, and might question whether local deployment makes model misuse harder to monitor or regulate. The Middle Ground camp would probably focus on the competitive dynamics, noting that open-weight releases from a hardware incumbent like Nvidia carry different implications than similar releases from purely software-focused labs.

The details most worth watching going forward include the specific licensing terms Nvidia attaches to the Bolt and Nemotron 3.5 weights, benchmark comparisons against rival open-weight models from Meta and Mistral, and whether enterprise adoption shifts measurably toward self-hosted deployments in the quarters following this release.