Build Multimodal AI Agents with NVIDIA Nemotron 3.5 Super VL on Bitdeer AI Model Studio
Always-on agents are no longer just reading and writing text. They need to interpret a screenshot, parse a scanned invoice, read a chart buried in a PDF, watch a clip of video, or navigate a live application interface, then reason over what they've seen and take the next action. That mix of seeing, reasoning, and acting is exactly where a text-only model runs out of road: teams end up stitching together a separate vision model, a text model, and a tool-calling layer just to finish one workflow.
NVIDIA Nemotron 3.5 Super VL is built to close that gap. Nemotron 3.5 Super VL is available from Day 0 on Bitdeer AI Model Studio, giving developers API access to an open, multimodal model without managing the underlying infrastructure.
Meet NVIDIA Nemotron 3.5 Super VL
Nemotron 3.5 Super VL is an open multimodal model designed for agentic workflows involving text, image, documents, application interfaces, and video. It combines visual understanding with agentic reasoning and tool use in a single model, so a workflow doesn't have to be assembled out of separate vision and text models.
Built on a 120B-parameter Mixture-of-Experts (MoE) architecture with 12B active parameters, Super VL combines accuracy and efficiency for specialized, always-on agent tasks. Organizations can post-train it with NVIDIA NeMo using their own data, tools, policies, and workflows.
Key Specifications
What can you build?
Customer service agents. Build agents that understand customer-provided screenshots, product images, and documents, reason about the information, and generate tool calls to support service workflows.
Document intelligence. Extract and interpret information from invoices, reports, forms, and documents containing complex layouts, tables, charts, and images. Use the results to support structured data processing and downstream workflows.
Computer-use agents. Understand application state, navigate interfaces, and verify outcomes across visual workflows.
Video understanding. Extract key information from video to support search, content review, advertising, retail, and incident analysis.
Multimodal research. Search across websites, connect information from multiple sources, and verify results.
Why run Super VL on Bitdeer AI Model Studio
Building multimodal applications is challenging enough without managing model-serving infrastructure.
Bitdeer AI Model Studio provides serverless inference access to advanced foundation models through a unified API, allowing developers to focus on application development rather than GPU provisioning and model deployment.
With Super VL on Bitdeer AI Model Studio, teams can:
- Get started quickly: Access the model through an API without deploying and managing their own inference infrastructure.
- Build multimodal applications: Integrate visual understanding, reasoning, and tool calling into existing workflows.
- Simplify model integration: Use Bitdeer AI Model Studio as a unified interface for accessing supported models.
Build on enterprise infrastructure: Take advantage of Bitdeer AI’s cloud infrastructure and secure capabilities.
Bitdeer AI is an NVIDIA Cloud Partner and reports ISO/IEC 27001:2022 and SOC2 Type I & Type II certifications.
Get Started with Nemotron 3.5 Super VL
Super VL is available through Bitdeer AI Model Studio
- Sign in to Bitdeer AI Model Studio.
- Find NVIDIA Nemotron 3.5 Super VL in the model list.
- Generate your API key and start making API calls.
from openai import OpenAI
client = OpenAI(
api_key="<API_KEY>",
base_url="https://api-inference.bitdeer.ai/v1",
)
response = client.chat.completions.create(
model="nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B",
messages=[
{
"role": "user",
"content": [
{"type": "text", "text": "What is in this image? Describe it briefly."},
{"type": "image_url", "image_url": {"url": "https://raw.githubusercontent.com/scikit-image/scikit-image/ee0a7a3ebd9ac8c2602f40e55bc015a3c8a81ae8/src/skimage/data/chelsea.png"}},
],
}
],
stream=False,
)
print(response)
Use the supported inference API to send your first multimodal request.
Start with a practical example: provide an image of an order confirmation and ask Super VL to extract the order number, identify the displayed status, and determine whether additional information is needed.
For agentic applications, you can connect the model's tool calls to your own services and business logic.
Explore NVIDIA Nemotron 3.5 Super VL on Bitdeer AI Model Studio.