You can run Nemotron 3.5 Super VL in LM Studio with a few clicks — it’s listed in LM Studio’s model catalogue with vision support and up to a 1M-token context. The smallest version is a Q4_K_M GGUF at about 70GB, so you need at least 70GB of RAM. My video below covers the model. Here’s how to download it, what you need, and how to plug it into your agents.
Key takeaways
- LM Studio offers a Q4_K_M GGUF of about 70GB.
- You need at least 70GB of RAM, plus extra for images and context.
- Supports vision: documents, charts, GUIs and visual Q&A.
- Connect LM Studio’s local server to Hermes as a custom endpoint.
How to get Nemotron 3.5 Super VL in LM Studio
- Install LM Studio for Mac, Windows or Linux.
- Search for Nemotron 3.5 Super VL in the model catalogue.
- Download the Q4_K_M build (about 70GB).
- Load it and start chatting — drag in an image or document page to test vision.
Hardware you need
LM Studio says you need at least 70GB of RAM for the smallest version, plus extra for visual inputs, context and runtime buffers. In practice that means a high-memory Mac or a workstation. If your machine is smaller, use the free OpenRouter variant instead.
| Machine | Verdict |
|---|---|
| 16–32GB laptop | Too small |
| 64GB machine | Borderline — likely too small with images |
| 96GB+ Mac or workstation | Should work |
| DGX Spark | Designed for models like this |
Want this working in your business, not just bookmarked? Using free open models in your AI agents is exactly what we build together inside the AI Profit Boardroom — 3,400+ members, four live calls a week, and a full one-hour DeepSeek Harness course in the classroom.
Prefer it mapped 1-on-1 first? Book a free strategy session and we’ll plan it for your exact situation.
Join the AI Profit Boardroom →Claim Your Free SEO Strategy Session →
What it can do in LM Studio
- Answer questions about documents and screenshots.
- Interpret charts and tables.
- Understand app and website screenshots (GUI understanding).
- Reason step by step, with tool use when connected to an agent.
Connect LM Studio to Hermes Agent
Start LM Studio’s local server (it uses an OpenAI-compatible API, usually on port 1234). In Hermes, add a custom endpoint pointing at it with the model’s name, in a dedicated profile. Set a context of at least 64K for agent work. You can do the same in OpenCode or any agent that supports custom endpoints.
LM Studio vs Ollama for this model
Both work. LM Studio gives you a friendly app with a model catalogue and chat window; Ollama gives you a simple command (ollama run nemotron-3.5-super) and an 80–90GB build. Pick whichever you already use — see my Ollama guide.
Let me be clear: this is not a frontier model. It’s not going to replace Claude Opus 5.5 for your hardest work. What it is: a free, open, fast model that can read screenshots, PDFs and charts and then act on them — which is exactly the job a lot of agent work starts with.
Getting good performance in LM Studio
Start with a modest context length and increase it only as needed — bigger contexts slow things down and use more memory. Offload as much of the model to your GPU as your hardware allows. Close other heavy apps before loading it. And use thinking mode only for tasks that need step-by-step reasoning, since it generates more tokens.
Using it for visual work
LM Studio lets you drag images into the chat. Try a screenshot of a spreadsheet and ask for the totals, a chart and ask what trend it shows, or a photo of a document and ask for a summary. For multi-page documents, send the most important pages as images. Clear, cropped images give the best results.
Private document work
The biggest reason to run it locally is privacy. Contracts, client reports, financial statements and medical letters can all be read and summarised without anything leaving your machine. For businesses that can’t send documents to cloud AI, a local vision model like this is a genuine option for the first time.
Final thoughts
If your machine can’t fit it, don’t worry — the free OpenRouter variant gives you the same model in the cloud, and LM Studio can still run smaller models locally for quick tasks. Many people keep a small local model for private, everyday work and use the cloud version of Nemotron for heavier visual jobs.
Either way, having a vision model in your toolkit that can read documents, charts and screenshots is a big upgrade for any agent setup.
How I’d test it this week
Before you download 70GB, check your free disk space and RAM, and test the cloud version on the same task first so you know what good looks like. If your machine qualifies, load it, try three private documents, and compare the answers with the cloud version.
If it works well, connect it to your agent as a private option for sensitive files.
Common questions
Does it work on a Mac? LM Studio runs on Mac, Windows and Linux; on a Mac you’ll need a high-memory model to fit the 70GB build. Is there a smaller version? LM Studio lists the Q4_K_M build as the smallest. Can it read PDFs? Send pages as images and it can read them. Is it as good as the cloud version? It’s the same model compressed to 4-bit, so expect very similar results with a small quality trade-off for much lower memory use.
The bottom line on Nemotron 3.5 Super VL LM Studio
Running Nemotron 3.5 Super VL in LM Studio is easy if you have 70GB+ of RAM: download, load and connect it to your agents for private, unlimited vision work.
FAQ: nemotron 3.5 super vl lm studio
Is Nemotron 3.5 Super VL in LM Studio?
Yes — it’s in LM Studio’s catalogue as a GGUF download.
How big is it in LM Studio?
The Q4_K_M build is about 70GB.
How much RAM do I need?
At least 70GB, plus extra for images and context.
Does the LM Studio version support images?
Yes — documents, charts, GUIs and visual Q&A.
Can I use it with Hermes?
Yes — connect LM Studio’s local server as a custom endpoint.
Can LM Studio run it on Windows?
LM Studio runs on Windows, Mac and Linux, but you still need at least 70GB of RAM for the smallest build.
Two ways I can help from here. Join the AI Profit Boardroom — it’s where your AI agent model setup gets built with 3,400+ members doing the same.
Or grab a free strategy session and bring your questions.
Join the AI Profit Boardroom →Claim Your Free SEO Strategy Session →
About Julian Goldie
I’m Julian Goldie — founder of Goldie Agency, best-selling author, and an AI educator with 428K+ YouTube subscribers. I’ve spent 10+ years in SEO and online business and I test every AI agent on my own work before I recommend it. Join the AI Profit Boardroom for the daily builds, or book a free strategy session to talk through yours.
Related reading
Last updated October 2026. This page is a living guide to nemotron 3.5 super vl lm studio — the facts here move fast and it is updated as they do.