Computers · mini PC
Framework Desktop
Hacker Approved
A compact, low-power mini PC with up to 128GB of shared memory for running large local AI models on Linux, though critics say memory bandwidth limits LLM speed.
Summarized from 163 comments.
A compact, low-power mini PC with up to 128GB of shared memory for running large local AI models on Linux, though critics say memory bandwidth limits LLM speed.
- recommend
- 93
- criticize
- 26
- months recommended
- 12 of 13
- median owned, 9 owners said
- 0.25 years
Why people love it
Runs large local AI models · 4
I use it exclusively as a server for LLM inference - first more for research but have been using it for coding now that there are sufficiently capable models that run on. For that I’m very happy.
Jul 2026 · commentI can comfortably run many even bigger oss models locally. You can dedicate 112GB to the GPU and it runs Linux perfectly.
Dec 2025 · commentI have a freedom of using a large library of various models and I can fit a lot in 128gb.
May 2026 · commentLower price for its memory · 4
you can still get a framework Max+ 395 with 128GB RAM for ~$2,459 USD. They have not increased the price for it yet.
Jan 2026 · commentYou can get a Framework Desktop with similar bandwidth for less than a third of the price
Nov 2025 · commentthe 128GB desktop mainboard being a bargain while their prices remain low rings true
Dec 2025 · commentWorks well with Linux · 3
I can comfortably run many even bigger oss models locally. You can dedicate 112GB to the GPU and it runs Linux perfectly.
Dec 2025 · commentit works flawlessly so far with Debian 13 install including (shock!) suspend working without issue
Nov 2025 · commentit's straightforward at least on Linux to have the GPU dynamically grab memory as needed
Apr 2026 · commentCompact and low power · 3
This thing compiles the linux kernel in 62 seconds. And it uses less power than my current machine to do it.
Mar 2026 · commentIt is noticeable faster, but what's better: the whole computer is tiny and sips energy.
Nov 2025 · commentA desktop computer in a compact form factor that can run AI workloads though is a much rarer thing.
Jul 2026 · comment
What people warn about
Memory bandwidth limits LLM speed · 4
Mac, DGX Spark, and a Framework Desktop / Ryzen AI Max 395 (ie Strix Halo) will not give you great performance running LLMs.
Jun 2026 · commentGiven that memory bandwidth remains the same I don't think it's worth buying if you have the 1st gen framework desktop.
Jul 2026 · commentUntil memory bandwidth significantly improves I just can't see myself wanting to use all that memory.
Jul 2026 · commentNoisy power supply · 2
my Framework Desktop definitely has the PSU noise issue, and it doesn't seem correlated to system load
Mar 2026 · commentThis case is cute and well built but it does have strange quirks, see also the noisy PSU
Mar 2026 · commentSoldered RAM and proprietary board · 2
it's still a shame Framework opted for a proprietary motherboard instead of mini-ITX with a socket CPU and GPU
Jul 2026 · commentShipping products with soldered RAM and CPU goes against that ethos.
Jul 2026 · commentHigh price for performance · 2
50% more RAM is not worth about 5% more performance for 3 times the price
Sep 2026 · commentthe Framework Desktop is indeed still at the (pretty inflated) price
Dec 2025 · comment
What hackers pick instead
- Apple Mac miniA small, fast, power-efficient Mac whose unified memory suits local AI models and always-on macOS work, though storage upgrades are costly and it struggles with big LLMs.390recommend
- NVIDIA DGX SparkA quiet, compact 128GB box with fast prompt processing for running large models locally, fine-tuning and keeping data private, though its low memory bandwidth slows token generation.170recommend
- AMD Strix HaloA normal x86 PC with up to 128GB of unified memory for running fairly large local models, though memory bandwidth limits speed, ROCm is fiddly and prices have climbed.130recommend
- Intel NUCSmall, quiet, cheap mini PC people use for home servers and basic desktop work, with replaceable RAM and storage. Critics say some models overheat.42recommend
- Beelink mini PCSmall mini PCs that run Linux or Windows well, stay quiet and cool, and last for years. People also praise the support and the value for the money.24recommend
Every quote
The 40 most detailed of 163 comments. Each links to the original.
I use it exclusively as a server for LLM inference - first more for research but have been using it for coding now that there are sufficiently capable models that run on. For that I’m very happy.
Recommends · Jul 2026 · in “New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory”
I can comfortably run many even bigger oss models locally. You can dedicate 112GB to the GPU and it runs Linux perfectly.
Recommends · Dec 2025 · in “Local AI is driving the biggest change in laptops in decades”
you can still get a framework Max+ 395 with 128GB RAM for ~$2,459 USD. They have not increased the price for it yet.
Recommends · Jan 2026 · in “Trinity Large: American Open Source Foundation Model”
This thing compiles the linux kernel in 62 seconds. And it uses less power than my current machine to do it.
Recommends · Mar 2026 · in “MacBook M5 Pro and Qwen3.5 = Local AI Security System”
my Framework Desktop definitely has the PSU noise issue, and it doesn't seem correlated to system load
Mixed · Mar 2026 · in “AMD will bring its "Ryzen AI" processors to standard desktop PCs for first time”
A desktop computer in a compact form factor that can run AI workloads though is a much rarer thing.
Recommends · Jul 2026 · in “New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory”
it works flawlessly so far with Debian 13 install including (shock!) suspend working without issue
Recommends · Nov 2025 · in “Framework Computer Now Sponsoring LVFS / Fwupd Development”
I have a freedom of using a large library of various models and I can fit a lot in 128gb.
Mixed · May 2026 · in “Memory has grown to nearly two-thirds of AI chip component costs”
You can get a Framework Desktop with similar bandwidth for less than a third of the price
Recommends · Nov 2025 · in “Asus Ascent GX10”
This case is cute and well built but it does have strange quirks, see also the noisy PSU
Mixed · Mar 2026 · in “AMD will bring its "Ryzen AI" processors to standard desktop PCs for first time”
it's straightforward at least on Linux to have the GPU dynamically grab memory as needed
Recommends · Apr 2026 · in “Framework Laptop 13 Pro”
It's half the price, and though I bought it for even less before RAM prices went crazy.
Recommends · May 2026 · in “Running local models on an M4 with 24GB memory”
It is noticeable faster, but what's better: the whole computer is tiny and sips energy.
Recommends · Nov 2025 · in “DRAM prices are spiking, but I don't trust the industry's why”
That's what I'm running on a Framework Desktop and I get ~50 tok/s plus or minus a few.
Recommends · May 2026 · in “Qwen3.7-Max: The Agent Frontier”
the 128GB desktop mainboard being a bargain while their prices remain low rings true
Recommends · Dec 2025 · in “Framework Raises DDR5 Memory Prices by 50% for DIY Laptops”
Only for chat sessions, not for agentic coding. It's just too slow to be practical
Mixed · Apr 2026 · in “Running Google Gemma 4 Locally with LM Studio's New Headless CLI and Claude Code”
Mac, DGX Spark, and a Framework Desktop / Ryzen AI Max 395 (ie Strix Halo) will not give you great performance running LLMs.
Criticizes · Jun 2026 · in “Two Qwen3 models on one DGX Spark: the residency math”
Given that memory bandwidth remains the same I don't think it's worth buying if you have the 1st gen framework desktop.
Criticizes · Jul 2026 · in “New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory”
it's still a shame Framework opted for a proprietary motherboard instead of mini-ITX with a socket CPU and GPU
Criticizes · Jul 2026 · in “New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory”
Until memory bandwidth significantly improves I just can't see myself wanting to use all that memory.
Criticizes · Jul 2026 · in “New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory”
50% more RAM is not worth about 5% more performance for 3 times the price
Criticizes · Sep 2026 · in “Framework Desktop with 192GB RAM”
the Framework Desktop is indeed still at the (pretty inflated) price
Criticizes · Dec 2025 · in “As AI gobbles up chips, prices for devices may rise”
Shipping products with soldered RAM and CPU goes against that ethos.
Criticizes · Jul 2026 · in “New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory”
spending $2k on a framework desktop one time to use as a k8s node means I have a much, much larger cluster than I'd be willing to pay for on a month to month basis
Recommends · Jan 2026 · in “Claude Code On-the-Go”
The framework desktop also has a small advantage where you can buy it as a standard ITX board and purchasing it that way gives you access to a 4x PCIe slot.
Recommends · Jul 2026 · in “Hands-On with the AMD Ryzen AI Halo”
Framework apparently benchmarked GPT-OSS 120B (!) on the maxed out 395+ Desktop and reached a 38.0 tok/sec Generation Speed
Recommends · Nov 2025 · in “Asus Ascent GX10”
It's definitely usable, but sometimes you're waiting a bit, though I'm not finding it problematic for the most part.
Recommends · Apr 2026 · in “Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model”
My Framework Desktop with 128GB was about half that. I did luck out by buying before RAM prices went crazy, though.
Recommends · Jun 2026 · in “Gemma 4 12B: A unified, encoder-free multimodal model”
bought framework desktop for a few thousand CAD back when everyone was calling framework desktop overpriced
Recommends · Jun 2026 · in “Qwen 3.6 27B is the sweet spot for local development”
For 800 eur I'd rather get the cheaper Framework desktop motherboard (Ryzen AI 385) and build around that.
Recommends · Nov 2025 · in “Steam Machine”
The main reason to get the framework is the DIY edition for the mini-itx form factor board and support.
Recommends · Apr 2026 · in “Google Gemma 4 Runs Natively on iPhone with Full Offline AI Inference”
Everything I play just works and the performance is good enough to just forget about as a concern.
Recommends · Nov 2025 · in “Linux gamers on Steam cross over the 3% mark”
If I get used to a Framework desktop running Linux then I'd probably stop buying MacBook Pros.
Recommends · Jun 2026 · in “Apple to Skip High-End M6 Mac Chips in Favor of AI-Focused M7 Line”
Gmktec used to be the cheapest but with the rising RAM prices its Framework noe i think.
Recommends · Feb 2026 · in “Qwen3.5: Towards Native Multimodal Agents”
I bought Framework Desktop in November 2025 with almost these exact specs for ~$2.5k
Recommends · Jul 2026 · in “AMD Ryzen AI Halo – $4k AI Dev Kit”
128gb is the max RAM that the current Strix Halo supports with ~250GB/s of bandwidth
Mixed · Mar 2026 · in “Apple discontinues the Mac Pro with no plans for future hardware”
If I were to upgrade, I would buy one of these: Framework desktop with 128gb for $3k
Recommends · Apr 2026 · in “GLM-5.1: Towards Long-Horizon Tasks”
A Mac, a Framework desktop, any number of mini PCs with some kind of unified memory.
Recommends · Nov 2025 · in “Gemini 3”
This is the type of system that is the a indirect byproduct of the change towards AI
Recommends · Nov 2025 · in “Samsung's 60% DRAM price hike signals a new phase of global memory tightening”
a framework desktop and omarchy is conceptually 98% of what the c100 purports to be
Recommends · Nov 2025 · in “C100 Developer Terminal”
Counts from 1,795,640 Hacker News comments, Nov 2025 to Nov 2026. How we count · Report a mistake