Running AI locally: what computer do you really need?

This article lays out the specifications for businesses in Belgium and the Netherlands: how much memory modern models really need, which shared-memory computers are available, what RAM and graphics cards currently cost, and when a hybrid approach is smarter than forcing everything to run locally.

What is a local AI computer?

A local AI computer runs a model in your own environment rather than on a model provider's servers. It could be a PC with a discrete graphics card, a compact shared-memory computer, or a Mac with unified memory.

With fully local processing, your prompts and files do not need to go to an external model provider. But local does not automatically guarantee privacy: an agent can still use an external API or web search. You therefore decide which connections are allowed. The Ollama documentation, for example, explains how to disable cloud features entirely.

Local inference does not incur per-token charges from a model provider. Hardware, electricity, updates, and support still cost money. And more parameters do not automatically mean a better answer: model architecture, training, and the specific task matter too.

The specifications that actually matter

System RAM has been underestimated in recent years. The old 32 GB guideline leaves little headroom if you want to run a larger model, an agent, a browser, and document processing at the same time. When buying, do not just choose an amount of RAM: check the memory budget for the exact model variant in the table below. Quantization, context length, and your offload approach together determine what you need.

With a discrete GPU, system RAM and video memory are separate. A PC with 64 GB of RAM and a graphics card with 16 GB of VRAM does not suddenly have 80 GB of fast GPU memory. If many model computations move to the CPU because the weights do not fit in VRAM, performance can drop considerably. Targeted offload of an n-gram lookup table is different: the whole table does not need to be processed continuously. Use ollama ps to check where Ollama has actually loaded your model.

Quantization reduces model weights by storing numbers at lower precision. The rule of thumb of roughly half a GB per billion parameters at 4-bit applies only to those weights. Scales, non-quantized components, image processing, temporary buffers, and the context cache add to that. In the session below, ollama ps reports 18 GB of memory use with a context setting of 32,768 tokens. That is one specific setup, not a fixed ratio between download size and memory use. Even with a MoE model, you must store all experts, not just the active ones. And a large context window on a model page does not mean you can use it for free: more context requires more memory.

Also choose fast NVMe storage with room for several models: large model packages can now require hundreds of GB each.

Which local models can you realistically run?

The selection below combines widely used families from the Ollama library with recent open models. It is not a ranking of the smartest AI. Model names link to Hugging Face. Package sizes come from the listed Ollama variants; for Qwen3.8-Flash-Next, from the Atomic Dynamic builds; and for GLM-5.3-Flash, from the Unsloth GGUF files (checked September 7, 2026). The memory advice applies to those specific builds. Download size is not the same as memory use: offload keeps part of the model on the SSD, while context, image processing, and parallel tasks require extra memory.

For Qwen3.8-Flash-Next, the package is larger than its label suggests. Alongside the 125B main model with 6B active parameters, the package includes an n-gram lookup table and an MTP component, bringing the total to about 180B parameters. But not everything needs to stay in fast memory permanently. The Qwen README explicitly describes the n-gram table as suitable for offload to host memory, with asynchronous prefetching. With suitable runtimes, that table can also stay on NVMe, with only the required portions read into memory.

The Atomic Chat guide documents a small Atomic Dynamic build on a Mac with 64 GB of unified memory, keeping the n-gram table on SSD. So 96 GB is not necessarily too little. This does require suitable quantization and a runtime with n-gram offload; it is not a promise that every Ollama or LM Studio package works this way.

End users report roughly 22–40 tokens per second during text generation with NVMe offload on different DGX Spark setups. The performance loss is limited in those tests, but the measurements in Hugging Face discussion #11 also show that prompt processing can become slower. Hardware, quantization, context, and runtime make a difference. Fully in-memory operation is generally faster and simpler, especially for long prompts; with suitable quantization, 128 GB is a practical starting point.

Open research is also exploring the limits of that memory budget: expertpin is a public experimental runtime research project that keeps MoE experts on NVMe and investigates how much fast memory such a model really needs. Its public target, at least 20 tokens per second within a tight RAM budget, is a development goal, not an achieved result.

GLM-5.3-Flash from Z.ai is even larger than Qwen3.8-Flash-Next and gpt-oss:120b, but can run entirely locally with enough memory. The multimodal model has an MIT license and a context window of about 1 million tokens. Unsloth's dynamic 4-bit variant, UD-Q4_K_XL, is 199.7 GB; for that variant, we recommend a computer with 256 GB of memory. Smaller dynamic quants can run on 128 GB, with lower precision and limited room for context. The 1-bit variant UD-IQ1_S, for example, is 93.1 GB. That is just the model package: the full context window does not automatically fit alongside the weights. Be careful with the Ollama model page: it only offers glm-5.3-flash:cloud, hosted by Ollama in the US and Europe, not on your computer. For local use, run Unsloth GGUFs through Unsloth Desktop or the appropriate llama.cpp build from the Unsloth guide.

Want to try it yourself? Ollama runs a model with a single command, and LM Studio provides a graphical environment for trying models side by side. Both tools are free. The terminal window below shows a real historical session from our own test machine, with the original Dutch prompt and response preserved: we load Qwen3.8 in its 27B variant with 4-bit compression, ask one question, and then use ollama ps to check where the model was loaded.

AI computers with unified memory

A discrete graphics card is no longer the only option. With unified or shared memory, the CPU and GPU use the same physical memory pool, so a compact computer can load models that do not fit in an ordinary graphics card's VRAM. But capacity is not speed: memory bandwidth, GPU compute, and software support still matter. A computer that barely fits the model is not automatically a comfortable work environment for a whole team.

Three families dominate this segment. The NVIDIA DGX Spark combines the GB10 Grace Blackwell Superchip with 128 GB of coherent LPDDR5x memory (273 GB/s) and a 4 TB NVMe SSD. NVIDIA explicitly positions the device for models up to 200 billion parameters. When we checked on September 7, the Founders Edition was in stock at Alternate in the Netherlands for € 6,399. Note that the system runs on Arm, so check compatibility with your other software.

The Mac Studio received a new generation on August 25, with M5 Max (up to 128 GB of unified memory) and M5 Ultra (up to 512 GB). Apple explicitly describes it as a desktop for on-device AI that runs enormous language models entirely locally. The Apple Store in Belgium and the Netherlands lists starting prices of € 3,029 (M5 Max) and € 6,649 (M5 Ultra); according to Apple, deliveries start September 22, with the 512 GB option following in late October. Those are base configurations, not maximum-spec AI configurations.

Then there is the AMD Strix Halo family: the Ryzen AI Max+ 395 combines 16 Zen 5 cores with Radeon 8060S graphics and supports up to 128 GB of unified memory. According to AMD, up to 96 GB can serve as graphics memory through Variable Graphics Memory. Compact mini PCs such as the GMKtec EVO-X2 with 128 GB cost around € 3,360. The LPDDR5x memory is soldered and cannot be upgraded later, so choose the capacity when buying. Above this segment is the DGX Station with 748 GB of memory (252 GB of HBM3e at the GPU plus 496 GB of LPDDR5X at the CPU). That is a different investment category, for which you request a quote based on your own workload.

RAM and GPU prices change your buying decision

Anyone planning an AI computer today faces a market-wide issue: memory is becoming scarce and expensive because data centers are absorbing supply. TrendForce reported that conventional DRAM contract prices in the first quarter of 2026 were about 93 to 98 percent higher than in the previous quarter, and forecast another increase of 58 to 63 percent for the second quarter. Industry demand directs suppliers toward high-capacity server modules for AI servers, leaving less capacity for ordinary PC memory. Those are industrial contract figures, not the retail price of your DDR5 kit, but the Tom's Hardware RAM price tracker shows consumer modules following the same curve.

Graphics cards follow the same pattern. The TechSpot GPU pricing survey from late August 2026 found that card prices rose by about 15 percent in a single month across ten countries, with GeForce cards taking the biggest hits. The original suggested retail price is therefore no longer a reliable budget: compare complete quotes and include RAM, storage, the power supply, cooling, installation, and support. In this market, a unified-memory computer with fixed memory can sometimes cost less than an expandable PC that you need to upgrade immediately.