Nemotron 3.5 Super VL is an open multimodal model from NVIDIA. It is a mixture-of-experts model with 120B total parameters, of which 12B are active for each token, so it runs with the compute cost of a much smaller dense model. It takes text, images, documents, GUI screenshots and video as input and answers in text, with optional reasoning, tool calling and structured outputs. It was released on 15 October 2026 and is available on AI/ML API as nvidia/nemotron-3.5-super-vl.
Nemotron 3.5 Super VL is the vision-language model of the Nemotron 3.5 generation. Nemotron 3.5 Lightning is a smaller text model (3B active out of 30B) built for high-throughput agentic workloads, and Nemotron 3 Super 120B-A12B is the earlier text-only model of the same size class. Super VL keeps the 120B/12B mixture-of-experts layout and adds visual input.
The input types point to three groups of tasks: document understanding (scanned pages, forms, charts and tables), GUI agents that read screenshots of an application before deciding on the next action, and video understanding. Tool calling and structured outputs let the model return its findings as JSON that downstream code can consume directly.
Input is billed at $0.34385 per 1M tokens, cached input at $0.13754 per 1M tokens and output at $2.0631 per 1M tokens. Reasoning tokens are billed at the output rate.
Document intelligence: extracting fields from invoices, forms and contracts, reading charts and tables, and answering questions about long scanned documents.
GUI and computer-use agents that take a screenshot of an application or web page, identify the elements on it and decide which action to take next through tool calls.
Video understanding, such as summarising a recording, finding the moment something happens or describing the steps shown in a screen capture.
Multimodal pipelines that need structured output, where images or video go in and a JSON object that follows a schema comes out.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "nvidia/nemotron-3.5-super-vl", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "nvidia/nemotron-3.5-super-vl", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"nvidia/nemotron-3.5-super-vl","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
Billed per token: $0.34385 / 1M input tokens, $0.13754 / 1M cached input tokens and $2.0631 / 1M output tokens. Reasoning tokens are billed as output.
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Intelligence | 26.5 | Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial Analysis | Source | October 16, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Nemotron 3.5 Super VL This page | Documents, GUI screenshots and video | |||
| High-throughput agentic workloads | ||||
| Coding + agents |
Yes. Nemotron 3.5 Super VL is live on AI/ML API as nvidia/nemotron-3.5-super-vl, through the chat completions endpoint.
It is a mixture-of-experts model with 120B total parameters, of which 12B are active for each token.
Text, images, documents, GUI screenshots and video. The model answers in text and supports structured outputs and tool calling.
The context window is 1,048,576 tokens, with up to 235,929 output tokens per request.
Nemotron 3.5 Lightning is a smaller text-only model (3B active out of 30B) aimed at high-throughput agentic work. Nemotron 3.5 Super VL is larger (12B active out of 120B) and reads images, documents, screenshots and video.
Nemotron 3.5 Super VL is priced at input $0.34385 / 1M tokens, cached input $0.13754 / 1M tokens and output $2.0631 / 1M tokens.
