Minisforum AI X1 Pro

[Enquiries]

Before I buy this system, I need to know for sure that other LLMs will run fully compatible - and your article says: - ""Only Features Microsoft Co-Pilot out the box – Hard/impossible to easily switch to ChatGPT or DeepSeek etc"" - if you do not know about this, please give me a contact to find out for sure... thank you

[jeremyclark]

Great question! I totally get wanting to nail down compatibility before dropping cash on a system like the Minisforum AI X1 Pro. The line you quoted—about it only featuring Microsoft Copilot out of the box and being hard to switch to other LLMs like ChatGPT or DeepSeek—seems to come from a review or article, right? I’ll break this down based on what I know about the X1 Pro and how LLMs typically work on hardware like this.
First off, the “Microsoft Copilot out the box” bit just means it’s preconfigured with Windows 11 and has Copilot integrated (it’s even got a dedicated Copilot key). That’s a marketing thing tied to the AMD Ryzen AI 9 HX 370 processor and Microsoft’s push for AI PCs. It doesn’t mean the system is locked to Copilot or can’t run other LLMs—it’s more about what’s ready to go when you boot it up.
Now, about running other large language models like ChatGPT or DeepSeek: the X1 Pro is a beefy little machine (up to 96GB RAM, NVMe SSDs, and that Ryzen AI chip with an NPU). Hardware-wise, it’s more than capable of handling local LLMs, including open-source ones like DeepSeek-R1 or distilled models (e.g., 7B or 14B parameter versions). The catch isn’t the hardware—it’s the software setup and how you want to use them.
Here’s the deal:

• ChatGPT: This is a cloud-based model from OpenAI, so you’d access it via API or browser, not run it locally. The X1 Pro’s Wi-Fi 7 and dual 2.5Gb Ethernet mean you’ll have no trouble connecting to it, but there’s no “local ChatGPT” to install since OpenAI doesn’t release its weights. Compatibility isn’t an issue here—it’s just not a local LLM.
  
• DeepSeek: This is where it gets interesting. DeepSeek’s models (like R1 or the distilled Qwen/Llama versions) are open-source and can run locally. With 64GB or 96GB RAM configs, the X1 Pro could handle something like DeepSeek-R1 14B (needs ~14GB RAM) or even 32B with some optimization (quantization, etc.). You’d just need to set it up yourself—install something like Ollama or LM Studio, grab the model files from DeepSeek’s repo, and tweak it to use the CPU/GPU/NPU. The NPU support might be tricky since AMD’s AI accelerators are still catching up to CUDA in terms of software ecosystem, but it’s doable with some elbow grease.
  

The “hard/impossible to switch” part from that article probably means there’s no built-in, user-friendly way to swap out Copilot for another LLM without extra setup. Out of the box, it’s optimized for Copilot, and Minisforum isn’t shipping it with a DeepSeek installer or anything. But that’s not a hardware limitation—it’s just a reflection of the default software stack. You’re not locked in; you just need to roll up your sleeves a bit.

[ed]

Hi, thanks for your message! To run other LLMs like ChatGPT or DeepSeek locally, you'll need a decent dedicated GPU (e.g. RTX 3060 or better), while CPU power is less critical for most models.