NVIDIA Nemotron 3.5 Super VL is free on OpenRouter with 1M context and multimodal input
OpenRouter added a free version of NVIDIA Nemotron 3.5 Super VL. Input and output cost nothing, with rate limits on the free endpoint. It reads images, document screenshots, and video, has about 1M tokens of context, and supports tool calls and structured outputs.
The briefing
OpenRouter launched a free version of NVIDIA Nemotron 3.5 Super VL on October 15, with model id nvidia/nemotron-3.5-super-vl:free. Input and output are billed at $0, the free endpoint is rate limited, and no end date has been announced.
What you can use
Nemotron 3.5 Super VL is an open multimodal mixture-of-experts model from NVIDIA, with about 120B total parameters and about 12B active. It takes text, images, documents, GUI screenshots, and video as input and returns text. The context window is about 1M tokens, with up to about 131K output tokens.
Reasoning is on by default; you can turn it off or give it a token budget. It supports tools for function calling and response_format for structured outputs. Good fits include document intelligence, web research, coding validation, customer service, video understanding, and long-running agents that combine vision with reasoning.
The paid nvidia/nemotron-3.5-super-vl is a separate, metered endpoint, so keep the :free suffix in your requests.
How to use it
- Sign in to OpenRouter and create an API key.
- In any OpenAI-compatible client or SDK, set the base URL to
https://openrouter.ai/api/v1. - Set the model to
nvidia/nemotron-3.5-super-vl:freeand send images or video using the multimodal message format.
Coding tools such as Kilo Code and Cline can also connect to OpenRouter and use this model. See the OpenRouter model page. If you only need a text model, the free Nemotron 3.5 Lightning from the same family is worth a look.