# BUYING ADVICE for local LLM machine
**URL:**
**Author:**  [@TheDefiled](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FTheDefiled)%5C
**Post date:** [March 24, 2025, 10:45am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F1 "2025-03-24T10:45:05Z")
Hy guys,
i want to buy/build a dedicated machine for local LLM usage. My priority lies on quality and not speed, so i’ve looked into machines with the capability for lots of “unified memory”, rather than GPU systems with dedicated fast but small VRAM. My budget would be “the cheaper the better”. I’ve looked at the “Nvidia - DGX Spark” but i must say for “only” getting 128 GB LPDDR5x of unified memory the price is too high in my mind.
Thanks for you suggestions!
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [March 24, 2025, 11:11am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F2 "2025-03-24T11:11:27Z")
The first thing I should say is, **let’s go to HF Discord**. There are people from multiple countries and regions who are knowledgeable about generative AI, so it’s perfect for hardware, especially GPU and CPU purchase consultations.
Now, if you can limit the use to LLM, the latest AMD isn’t bad, but CUDA is also attractive…
The proprietary OS seems fast, but I’m worried about the breadth of use cases.
> **[$3,000 for THIS? NVIDIA’s DGX Spark Faces Tough Competition | Hardware Corner](https://www.hardware-corner.net/nvidias-dgx-spark-digits-specs-20250319/)**
>
> After months of speculation and anticipation, NVIDIA has finally unveiled the full specifications for its DGX Spark workstation (formerly known as Project DIGITS), aimed at AI developers and…
> **[Unsloth Requirements | Unsloth Documentation](https://docs.unsloth.ai/get-started/beginner-start-here/unsloth-requirements)**
>
> Here are Unsloth's requirements including system and GPU VRAM requirements.
---
**Author:**  [@TheDefiled](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FTheDefiled)%5C
**Post date:** [March 24, 2025, 5:54pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F3 "2025-03-24T17:54:19Z")
Thank you for your input, i will check out those sources!
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [March 24, 2025, 11:51pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F4 "2025-03-24T23:51:02Z")
Or rather, if you go to Discord, it’s really reliable.😅 If you’re buying something expensive, it’s often cheaper to ask someone who already has it. I haven’t bought a new environment for AI yet… I’m just watching from the side.
Good luck!
---
**Author:**  [@TheDefiled](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FTheDefiled)%5C
**Post date:** [March 25, 2025, 7:59am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F5 "2025-03-25T07:59:32Z")
I have a good Thread running on Reddit when you are interested:
> **[Reddit - The heart of the internet](https://www.reddit.com/r/LocalLLaMA/comments/1jinpyx/buying_advice_for_local_llm_machine/?rdt=38243)**
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [March 25, 2025, 9:50am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F6 "2025-03-25T09:50:33Z")
Oh, I didn’t know GAIA…
[https://www.amd.com/en/developer/resources/technical-articles/gaia-an-open-source-project-from-amd-for-running-local-llms-on-ryzen-ai.html](https://www.amd.com/en/developer/resources/technical-articles/gaia-an-open-source-project-from-amd-for-running-local-llms-on-ryzen-ai.html)
---
**Author:**  [@TheDefiled](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FTheDefiled)%5C
**Post date:** [March 25, 2025, 9:56am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F7 "2025-03-25T09:56:44Z")
It was new for me too!
---
**Author:**  [@Androphin](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FAndrophin)%5C
**Post date:** [March 25, 2025, 6:21pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F8 "2025-03-25T18:21:08Z")
I have a similar need, because I can’t use computing-clouds for tests with sensitive data from users.
And as far as I am right now with figuring out the amazing HF universe, it seems everything is uploaded/processed in the cloud.
For e.g. FastRTC seems coupled with HF, if a temporary telephone number can be acquired. To me it’s often not clear, where the data gets processed.
My interests are in Speech Synthesis for making apps assessable by voice.
Analyzing documents (PDF) for classification and summarization to build a sorting app.
But its hard for me to figure out what’s still possible on my local computer or if I can build such AI driven applications. Later wouldn’t be normally able to run on “ordinary” hardware, right?
Any hints from your experienced view?
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [March 26, 2025, 6:54am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F9 "2025-03-26T06:54:00Z")
I’m not very experienced, but I think it would be possible to achieve this with a GPU for private use if you connect an ASR model like Whisper that works locally with a tool that converts PDFs to markdown or text.
If you want to use FastRTC for confidential documents, you can set up your own server, but if you want to do it completely locally, it would probably be easier to use a local model such as Whisper from the start. These will be fine as long as you have 2GB of VRAM.
I think you can use any LLM or LLM to summarize. The appropriate model will depend on the purpose and the complexity of the documents you are working with, but most LLM can handle this task, even if it is small. LLM requires as much VRAM as you want, but if you just want to summarize, even a model with 5GB of VRAM in quantized state should not normally cause any problems. Well, if you have 12GB of VRAM to spare, you can achieve it. Well, it’s a mid-level GeForce.
For more information, please refer to the Hugging Face Audio Course, which explains how to create the framework for various things.
> **[Spaces - Hugging Face](https://huggingface.co/spaces?q=asr&sort=trending)**
>
> Discover amazing ML apps made by the community
> **[Spaces - Hugging Face](https://huggingface.co/spaces?q=pdf&sort=trending)**
>
> Discover amazing ML apps made by the community
> **[Creating a voice assistant - Hugging Face Audio Course](https://huggingface.co/learn/audio-course/chapter7/voice-assistant)**
>
> We’re on a journey to advance and democratize artificial intelligence through open source and open science.
### PDF Converters works also on locally (if models were already downloaded)
> **[Pdf 2 Summary - a Hugging Face Space by zamal](https://huggingface.co/spaces/zamal/Pdf-2-Summary)**
>
> Upload a PDF document and select the extent of summarization to get a concise bullet-point summary of its key points.
> **[PDF to Markdown - a Hugging Face Space by huggingchat](https://huggingface.co/spaces/huggingchat/pdf-to-markdown)**
>
> Upload a PDF to extract its text content and metadata. If the PDF contains images, it will use OCR to improve text extraction. You'll get the text in markdown format and the PDF's metadata as a JSO...
### Fast RTC Deployment
> **[Deployment - FastRTC](https://fastrtc.org/deployment/)**
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [March 26, 2025, 6:55am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F10 "2025-03-26T06:55:23Z")
For converting PDF to something, also refer this post.
> [@Any model that takes in a clean PDF and outputs a JSON of all the fillable fields that should be added to it + coordinates?](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fany-model-that-takes-in-a-clean-pdf-and-outputs-a-json-of-all-the-fillable-fields-that-should-be-added-to-it-coordinates%2F147198)%3A
>
> So basically I’m searching for a model that can take a clean PDF, and output a JSON of all the fillable fields, with relevant names, types, page number and most importantly, coordinates in the PDF file. What I mean by clean is just a PDF form without the fields embedded into it to allow filling. I already know how to add the fields to the PDF it self and save it, but I need something that will output me the fields to add and where to add them. Is there any model like this in HuggingFace? If you…
---
**Author:**  [@podgy87](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fpodgy87)%5C
**Post date:** [March 10, 2026, 4:53am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fbuying-advice-for-local-llm-machine%2F147169%2F11 "2026-03-10T04:53:51Z")
> [@Any model that takes in a clean PDF and outputs a JSON of all the fillable fields that should be added to it + coordinates?](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fany-model-that-takes-in-a-clean-pdf-and-outputs-a-json-of-all-the-fillable-fields-that-should-be-added-to-it-coordinates%2F147198%2F1)%3A
>
> n’t a lot at all online, see [https://www.youtube.com/watch?v=-SlgDIjVW1g](https://www.youtube.com/watch?v=-SlgDIjVW1g)
If you do manage to get everything humming along, I’d love to share something back with you. I put together a basic web interface that might make things a bit more accessible. You can find it here: [GitHub - geekie87/Aria-UI: ARIA — Adaptive Reasoning Intelligence Assistant · GitHub](https://github.com/geekie87/Aria-UI)