NVIDIA's Nemotron 3.5 Open-Model Family at GTC Berlin

Episode 1 · NVIDIA Unveils Nemotron 3.5 and Open-Weight Model Family at GTC Berlin (2026-10-20, 7 posts)

At GTC Berlin, NVIDIA announced a major expansion of its open model lineup, rolling out seven model families at once—including Nemotron 3.5—spanning language, vision, robotics, biology, and quantum computing. The release leans into Europe's "sovereign AI" narrative: enterprises and institutions can fine-tune on local data and deploy on their own controlled infrastructure, targeting developers, researchers, and businesses building specialized AI agents, training robots, designing proteins, and developing quantum applications.

Confirmed

• The release was officially announced during GTC Berlin, giving European teams more open-weight models
• Nemotron 3.5 Super VL: a multimodal (text and image) open language model with 120 billion parameters, aimed at enterprise-grade Agents and agent orchestration; smaller than Nemotron 3 Ultra and already available on Hugging Face
• The batch also includes Nemotron 3.5 Super, Nemotron Bolt, and Nemotron Retriever series, plus models for robotics, biology, and quantum computing
• Target use cases fall into two categories: persistent Agents and on-device computing (AI PC)

Why it matters

• This is NVIDIA's systematic play for European enterprises' "sovereign AI" needs: models customizable with local data and deployable on local infrastructure, directly addressing Europe's demands for data and compute autonomy
• The open-weight strategy lowers the barrier for enterprises to build specialized Agents and train robots, while strengthening NVIDIA's presence in the model ecosystem (beyond just GPUs)
• The multimodal + retrieval + on-device combination signals NVIDIA is targeting the full pipeline from cloud-based enterprise Agents to AI PCs

• Nvidia launches open Nemotron 3.5 Super VL and Bolt models for enterprise agents and AI PCs — emmanuelvivier · 2026-10-20
• NVIDIA releases seven open model families at GTC Berlin, from Nemotron 3.5 to Cosmos 3.5 and BioNeMo — NVIDIAAI · 2026-10-21
• NVIDIA launches Nemotron 3.5, Cosmos 3.5 and more open models for Europe's sovereign AI push — testingcatalog · 2026-10-21
• NVIDIA GTC Berlin open-model drop: Nemotron 3.5 Super VL, Retriever, RFD4-Proteina — AllThingsApx · 2026-10-21
• NVIDIA announces more open models for Europe at GTC Berlin for agents, robots and protein design — DesireeCachette · 2026-10-21
• NVIDIA unveils model portfolio spanning language, robotics, biology and quantum — AllThingsApx · 2026-10-21
• NVIDIA drops 8 open model families at GTC Berlin, pitching sovereign AI for Europe — AllThingsApx · 2026-10-21

Episode 2 · NVIDIA Releases Nemotron 3.5 Retriever Family on Baseten (2026-10-21, 2 posts)

NVIDIA expanded its Nemotron Retriever lineup with open, commercially available models for enterprise agent workflows — Nemotron 3.5 Embed, Parse, and Rerank — now available on Baseten's inference platform.

• NVIDIA Nemotron Retriever models (Embed, Parse, Rerank) land on Baseten — baseten · 2026-10-21
• NVIDIA ships Nemotron 3.5 Embed, Rerank, and Parse open retrieval models for agents — baseten · 2026-10-21