Nemotron 3.5 Super VL
Neural Network
Nemotron 3.5 Super VL is an open NVIDIA multimodal model for analyzing images, documents, and video.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 1 048 576 tokens
- Max answer length
- 943 718 tokens
- Reasoning effort
- low, medium, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Nemotron 3.5 Super VL?
Nemotron 3.5 Super VL is an open NVIDIA multimodal model based on mixture-of-experts architecture: out of 120B parameters, 12B are active per request. It accepts text, images, documents, interface screenshots, and video, allowing the model to analyze both the task description and surrounding visual material. Via BotHub, it is accessible from Russia without a VPN or foreign card; payment is made with Russian cards in rubles, charged per token without a subscription. Over 250 models are available in one window, making it easy to compare answers and switch. A unified OpenAI-compatible API allows connecting Nemotron to your services without rewriting integrations. Dialogues are encrypted via AES-GCM, data is not stored, and companies have access to contracts, invoices, EDI, and an admin panel with limits. Simple scenarios include: extracting data from scans and tables, describing frame or photo content, analyzing interface screenshots for QA and agent tasks, summarizing video recordings, and building multimodal pipelines where text and images are processed in a single request.Frequently asked questions about Nemotron 3.5 Super VL
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
nemotron-3.5-super-vl is an NVIDIA model that works with text, documents, and images. It reasons before answering and supports function calling, making it suitable for analyzing long documents, coding, math and logic tasks, and scenarios involving external tools.
The output limit is 235,929 tokens per response. This is enough for a voluminous technical document, major refactoring, or translating an entire chapter without splitting. The context window is 1,048,576 tokens, so input materials can be uploaded in their entirety.
Yes, reasoning before answering is confirmed by our data. The model breaks down the task step-by-step before formulating the result, which significantly helps with math, logic, code debugging, and tasks where a chain of thought is more important than a quick response.
Yes, both function calling and structured JSON output are supported. You describe the schema and available tools, and the model returns the answer in the specified format—this is convenient for agents, data parsing, and integrations. In BotHub, access is provided via a unified OpenAI-compatible API.
Images and documents—yes: you can send a screenshot, diagram, chart, or PDF and ask to analyze the content. Audio and video are not listed as supported inputs in our data, so it is better to choose a specialized model in BotHub for those.
We do not have data on prompt and response languages, so we cannot guarantee the quality of Russian. Test it with your task: in BotHub, you pay per token, and you can switch to another model in the same window.