16GB max RAM · 2 bays · 2 x 2.5GbE
AINAS Score 31
£479
Amazon UK · Verified

AI CapableIceWhale
Why AI Capable: Intel N100 with no NPU listed, a Gen3 PCIe slot with no documented GPU support on this model, and 16GB maximum memory, so it runs containers and small CPU models only.
Updated
Can the IceWhale ZimaCube run local AI?
The IceWhale ZimaCube can run small local models, with limits. Its Intel N100 and 16GB memory ceiling suit 7B to 8B class models (Estimated), so treat it as a storage-first NAS that can host light AI tools.
There is no way to add a discrete GPU, so all inference runs on the CPU.
Local LLM support is not yet verified.
The ZimaCube suits a buyer who wants a compact six-bay home server with fast flash storage and container support, and who treats AI as a small extra. It has 6 bays and 5 M.2 slots, two 2.5GbE ports, an Intel N100 with 4 cores, and 8GB of DDR4 3200 SODIMM memory that can be taken to 16GB. It runs ZimaOS with Docker and virtual machine support, and the PSU is 220W. On model fit, Qwen3 8B fits at Q4 (Estimated), and Llama 3.1 8B and Qwen3 8B are marginal at Q8 (Estimated). Mistral Small 3 24B does not fit at Q4 (Estimated). The AINAS Score is 31. Treat it as a home lab and storage machine first, with small CPU models as a bonus.
The 16GB memory ceiling and the missing GPU path rule the ZimaCube out for anyone who wants to run local LLMs seriously. Memory is a single SODIMM, and the PCIe Gen3 slot has no documented GPU support on this model. IceWhale's own LLM guide asks for a GPU with 16GB or more VRAM, and the local LLM capability here is Unverified. Gemma 3 12B does not fit at Q8 (Estimated), and Mistral Small 3 24B does not fit at Q4 or Q8 (Estimated). Buyers who need larger models should look for a system with a documented GPU slot, an NPU, or a much higher memory ceiling. Treat every AI feature on the ZimaCube as unconfirmed until it is verified.
No verified price
We have not verified a listing for the IceWhale ZimaCube at a retailer we track. The specifications below stand on their own; a price appears here as soon as a listing is confirmed.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon.co.uk or Amazon.com, as applicable, at the time of purchase will apply to the purchase of this product.
Each capability shows how it was confirmed: Verified by AINAS, Vendor confirmed, Community confirmed or Unverified.
| Capability | Supported | Evidence | Note |
|---|---|---|---|
| Local LLM | Limited | Unverified | IceWhale LLM guide requires a GPU with 16GB or more VRAM; N100 with 16GB would run small CPU models at most. |
| Ollama | Limited | Unverified | Same ZimaOS stack; no evidence on the N100 model. |
| Open WebUI | Limited | Unverified | Installable from the ZimaOS App Store; IceWhale guide is written for the Pro Creator Pack. |
| RAG | Limited | Unverified | No evidence on this model. |
| AI search | Limited | Unverified | Page describes AI Assist for the ZimaCube range without separating models. |
| GPU expansion | Limited | Unverified | Gen3 slot of about 2 GB/s; IceWhale names only the Pro in its GPU guide. |
| Hardware acceleration | Limited | Unverified | N100 integrated graphics not documented for acceleration on this page. |
| AI photos | Limited | Vendor confirmed | PhotoPrism appears in the App Center section. |
| AI video | No | Unverified | No AI video feature documented. |
| Surveillance AI | Limited | Unverified | Frigate tutorial exists for ZimaOS but needs a compatible GPU. |
Calculated from the ZimaCube's memory and each model's published size at each quantization. Rows marked Estimated are calculations, not benchmarks. Check any model.
| Model | Best quantization | Runs? | Memory needed | Expected performance | Verification |
|---|---|---|---|---|---|
| Llama 3.2 3BLlama, 3.21B | Q8 | Yes | 13GB | Not measured | Estimated |
| Llama 3.1 8BLlama, 8.03B | Q5 | Yes | 15GB | Not measured | Estimated |
| Qwen3 8BQwen, 8.19B | Q4 | Yes | 15GB | Not measured | Estimated |
| Gemma 3 12BGemma, 11.8B | Q4 | Marginal | 15GB | Not measured | Estimated |
| Qwen3 14BQwen, 14.8B | Q4 | Marginal | 16GB | Not measured | Estimated |
| Phi-4 14BPhi, 14.7B | Q4 | Marginal | 16GB | Not measured | Estimated |
| Mistral Small 3 24BMistral, 23.6B | None fits | No | 25GB at Q4 | Not measured | Estimated |
| Gemma 3 27BGemma, 27B | None fits | No | 29GB at Q4 | Not measured | Estimated |
| Qwen3 30B A3BQwen, 30.5B | None fits | No | 31GB at Q4 | Not measured | Estimated |
| Qwen3 32BQwen, 32.8B | None fits | No | 32GB at Q4 | Not measured | Estimated |
| DeepSeek-R1 32B (Qwen distill)DeepSeek, 32.8B | None fits | No | 32GB at Q4 | Not measured | Estimated |
| Llama 3.3 70BLlama, 70.6B | None fits | No | 60GB at Q4 | Not measured | Estimated |
| Llama 4 Scout 17Bx16ELlama, 109B | None fits | No | 89GB at Q4 | Not measured | Estimated |
| Qwen2.5 72BQwen, 72.7B | None fits | No | 65GB at Q4 | Not measured | Estimated |
| Mistral Large 2 123BMistral, 123B | None fits | No | 96GB at Q4 | Not measured | Estimated |
One verdict per workload, from the ZimaCube's use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
We cannot confirm local LLM use on the ZimaCube. The capability is marked limited and Unverified, and IceWhale's guide asks for a GPU with 16GB or more VRAM, while the N100 with 16GB would run small CPU models at most.
Private RAG is unknown for the ZimaCube. Open WebUI is installable from the ZimaOS App Center but is Unverified, and the RAG row has no evidence on this model, with IceWhale's guide written for the Pro Creator Pack.
AI search cannot be confirmed on the ZimaCube. IceWhale describes AI Assist for the ZimaCube range without separating models, and the row is limited and Unverified, so we cannot say it applies to this model.
AI photos is a poor fit for the ZimaCube. PhotoPrism appears in the App Center section and IceWhale confirms it as limited, but the N100 has no NPU and its integrated graphics are not documented for acceleration.
Surveillance is unknown on the ZimaCube. A Frigate tutorial exists for ZimaOS, but it needs a compatible GPU, the system has no documented GPU path, and the row is limited and Unverified, so nothing confirms it here.
Home lab use is OK on the ZimaCube. It runs Docker and virtual machines on ZimaOS, with an Intel N100 with 4 cores, 6 bays, 5 M.2 slots and two 2.5GbE ports, though memory stops at 16GB.
Business use is a poor fit for the ZimaCube. It has no ECC memory, memory is capped at 16GB in a single SODIMM, no warranty period is listed, and networking stops at two 2.5GbE ports on this model.
Creative AI is unknown on the ZimaCube. GPU expansion is limited and Unverified, with a Gen3 slot of about 2 GB/s and IceWhale naming only the Pro in its GPU guide, and hardware acceleration is also Unverified.
The ZimaCube ships with 8GB of DDR4 3200 SODIMM and supports up to 16GB. Upgrading to the maximum lifts the model ceiling from small models up to around 7B parameters to 7B to 8B class models (both Estimated).
As a rule of thumb, a 4-bit model needs about 0.6GB per billion parameters plus room for context. After leaving around 8GB for the NAS operating system, 16GB leaves space for a model of roughly 13B parameters at most (Estimated).
No. The ZimaCube has no PCIe slot or OCuLink port for a graphics card. Inference runs on the CPU, so keep to small models.
With limits. The ZimaCube can host a small RAG stack such as AnythingLLM or Open WebUI with a compact embedding model, but memory and compute restrict how large the answering model can be. It suits a personal document library more than a team knowledge base.
Networking is 2 x 2.5GbE. Fine for one or two users; indexing a large archive over the network will be slower.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| N100, 8GB, 256GB | 8GB | 256GB |
The closest IceWhale model by bay count and AINAS Score. Only the specifications that differ are listed.
The ZimaCube scores best for Home Lab (40/100) and Business (39/100). It suits home lab owners who want one box for storage, containers, media and a few AI services.
It is a weaker fit for anyone who wants to run models larger than about 8B parameters and for buyers who expect to add a graphics card later.
Other AI Capable systems closest in overall score.
Yes. Qwen3 8B fits at Q4 (Estimated) and is marginal at Q8 (Estimated), meaning it loads with little headroom. That is a calculation from memory, not a benchmark. The ZimaCube has 8GB of memory included and a 16GB maximum, and the local LLM capability is Unverified, so treat it as unconfirmed in practice. Mistral Small 3 24B does not fit at Q4 (Estimated).
Yes, up to 16GB. The ZimaCube takes DDR4 3200 SODIMM memory, ships with 8GB, and is listed with a single SODIMM at the maximum of 16GB. It does not support ECC memory. The ceiling matters for AI work, because Gemma 3 12B is only marginal at Q4 (Estimated) and Qwen3 14B is marginal at Q4 and does not fit at Q8 (Estimated).
No, not as a documented option. GPU expansion is listed as none for this model, and the Unverified GPU row describes a Gen3 slot of about 2 GB/s. IceWhale names only the Pro in its GPU guide. The ZimaCube also has no Thunderbolt, USB4 or OCuLink port. If you need a GPU for local models, look for a system with a documented GPU slot.
Yes, 5. The ZimaCube lists M.2 for the OS plus four on the 7th bay, alongside 6 drive bays. The count comes from the maker's specification table and should be checked against the maker's page before you buy. The configuration we list has 256GB of SSD storage and 8GB of memory. Two 2.5GbE ports carry network traffic to that storage, and there is no 10GbE port.
No. The Intel N100 has 4 cores and no NPU is listed, and IceWhale does not document its integrated graphics for acceleration on this model, so hardware acceleration is limited and Unverified. IceWhale's own LLM guide asks for a GPU with 16GB or more VRAM, and this model has no documented way to take one. That leaves small models on the CPU, such as Qwen3 8B at Q4 (Estimated), with local LLM support Unverified.
Yes. ZimaOS on the ZimaCube provides an App Center with Docker support, and a KVM environment for virtual machines. The N100 has 4 cores and 8GB of memory is included, which can be raised to 16GB, so general containers are practical. AI apps such as Open WebUI, installable from the App Center, are marked limited and Unverified.
Partly. Ollama support is not yet verified. See the capability table above for the source.
Partly. Open WebUI support is not yet verified. See the capability table above for the source.
Partly. AI photos support is documented by the manufacturer. See the capability table above for the source.
Partly. Surveillance AI support is not yet verified. See the capability table above for the source.
Partly. Hardware acceleration support is not yet verified. See the capability table above for the source.
It has 2 x 2.5GbE built in, and the PCIe Gen3 at 2 GB/s plus a Gen3 1 GB/s slot can take a faster network card. The fastest port is 2 x 2.5GbE.
Every field records its source. Specifications were last checked on . Spot an error? Send a correction with a manufacturer link.