NVIDIA releases open-weight Nemotron 3.5 Super VL for text and vision agents
Article timeline
The articles in this story. Times are ET.
fal@falOfficialfal offers Nvidia Nemotron 3.5 Super for text, vision, and videoAIfal links to three hosted endpoints for Nvidia's Nemotron 3.5 Super model, covering text, vision, and video. The post itself gives no benchmarks, pricing, or parameter counts.
OpenRouter@OpenRouterOfficialOpenRouter offers free access to NVIDIA Nemotron 3.5 Super VLAIOpenRouter links to a free model page for NVIDIA's Nemotron 3.5 Super VL at The post gives no further details on the model's capabilities, pricing terms, or limits.
LM Studio@lmstudioOfficialNemotron 3.5 Super VL is available in LM Studio for local useAILM Studio now offers NVIDIA's Nemotron 3.5 Super VL for local use. The model is a 120B total, 12B active mixture-of-experts model that accepts vision input and supports reasoning and tool use. Running it locally requires about 70GB of RAM.
- fal@falOfficialNVIDIA Nemotron 3.5 Super VL goes live on fal as an open-weight multimodal model
AIfal has made NVIDIA's Nemotron 3.5 Super VL available, an open-weight multimodal model with 120B total parameters and 12B active per token. NVIDIA says the model combines reasoning and tool use with text, image, document, and video understanding for long-running AI agents. fal lists visual reasoning, tool calling, and document intelligence among its use cases.
- fal@falOfficialfal offers Nvidia Nemotron 3.5 Super for text, vision, and video
AIfal links to three hosted endpoints for Nvidia's Nemotron 3.5 Super model, covering text, vision, and video. The post itself gives no benchmarks, pricing, or parameter counts.
LangChain@LangChainOfficialNVIDIA's Nemotron 3.5 Super VL runs in LangChain Deep AgentsAILangChain says NVIDIA's open-weight Nemotron 3.5 Super VL, a 120B-parameter model with 12B active, averaged an 87% pass rate on its Deep Agents behavioral suite. On the Harbor-lite benchmark, the model finished tasks in 55% less total time than DeepSeek V4.1 Flash at similar success, and it is accessible through ChatNVIDIA.
NVIDIA AI@NVIDIAAIOfficialTeams post-train Nemotron 3.5 Super VL across five enterprise tasksAINVIDIA says teams are already post-training Nemotron 3.5 Super VL for document intelligence, cybersecurity, finance, scientific experimentation, and travel. A chart in the post shows accuracy rising as the model is post-trained for each task.
- NVIDIA AI@NVIDIAAIOfficialNemotron 3.5 Super VL combines text and visual understanding in one model
AINVIDIA says Nemotron 3.5 Super VL unifies text and visual understanding in a single model for agents that read documents, interpret charts, and search the web. This removes the need to pass context between separate text and vision models.
Together AI@togethercomputeOfficialTogether AI offers Nemotron 3.5 Super VL on its platformAITogether AI says Nvidia's Nemotron 3.5 Super VL is now available to run on its platform. The post links to a model page for developers to start building.
- Together AI@togethercomputeOfficialTogether AI hosts NVIDIA Nemotron 3.5 Super VL, an open-weight multimodal model
AITogether AI says NVIDIA Nemotron 3.5 Super VL is now live on its platform. The open-weight multimodal model is built for visual reasoning and always-on agents that process documents, screens, and video.
