64GB max RAM · 24 bays · 1GbE
AINAS Score 43
No verified price

AI CapableASUSTOR
Why AI Capable: Intel Atom C3538 (4 cores, no iGPU, no NPU) is capped at AI Capable by the Atom rule despite a 128GB memory ceiling; the one PCIe 3.0 slot (x4 link) has no documented GPU use.
Updated
Can the ASUSTOR Lockerstor 12RD (AS6512RD) run local AI?
The ASUSTOR Lockerstor 12RD (AS6512RD) can run small local models, with limits. Its Intel Atom C3538 and 128GB memory ceiling suit 70B class models at higher-quality quantization, or large mixture-of-experts models at 4-bit (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 documented by the manufacturer.
The Lockerstor 12RD suits a buyer who wants a 12-bay rackmount NAS that can also run small local AI jobs on the side. It has an Intel Atom C3538 with 4 cores, 8GB of DDR4-2133 SO-DIMM memory in a 4-slot layout, and a 128GB ceiling. Four network ports (2 x 2.5GbE, 2 x 1GbE), two redundant 350W supplies, Docker and virtual machine support, and a 3 year warranty cover the storage and home lab side. Ollama with Open WebUI is vendor confirmed on ADM 5. Mistral Large 2 123B fits at Q4 with 128GB installed (Estimated).
Inference is CPU-only, and that is the limit for anyone who bought this for AI. The Atom C3538 has no integrated GPU and no NPU, and ASUSTOR states its NAS models do not feature a discrete GPU. The single PCIe 3.0 slot runs at x4 and has no documented GPU use. ASUSTOR recommends an external GPU host for inference. Buyers who want fast model responses should look for a discrete GPU path, a PCIe x16 slot, OCuLink or Thunderbolt for an external GPU, or an NPU with a stated TOPS figure. The Lockerstor 12RD is better judged as storage that can hold models.
Amazon UK
8GB
£2,611.20
Verified
In stock
Buy on AmazonProduct 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.
Price history is building. We started tracking this listing on 13 November 2026 and have 23 verified prices so far. The chart and the averages switch on automatically after 7 days of history.
Each capability shows how it was confirmed: Verified by AINAS, Vendor confirmed, Community confirmed or Unverified.
| Capability | Supported | Evidence | Note |
|---|---|---|---|
| Local LLM | Limited | Vendor confirmed | ASUSTOR App Central lists Ollama + Open WebUI (v0.11.4.r01, ADM 5.0+) for the AS65 series. CPU-only inference on Intel Atom C3538. ASUSTOR recommends an external GPU host for inference. |
| Ollama | Yes | Vendor confirmed | Ollama + Open WebUI app lists AS65 as a supported series. |
| Open WebUI | Yes | Vendor confirmed | Bundled with the Ollama + Open WebUI App Central package. |
| RAG | Limited | Vendor confirmed | ASUSTOR markets NAS as RAG knowledge base storage with vector databases (Chroma, pgvector, Weaviate) via Docker. It recommends an external compute host for inference. |
| AI search | No | Unverified | No AI or semantic file search feature found in ADM 5.1 material. |
| GPU expansion | No | Vendor confirmed | ASUSTOR states its NAS models do not feature a discrete GPU. The one PCIe slot (x4 link) has no documented GPU use. |
| Hardware acceleration | No | Unverified | ASUSTOR spec lists multimedia output N/A and no GPU. No NPU. |
| AI photos | Yes | Vendor confirmed | Photo Gallery app lists AS65; ADM 5.1 adds facial recognition grouping powered by locally run AI. |
| AI video | No | Unverified | No AI video feature found in ADM documentation. |
| Surveillance AI | Yes | Vendor confirmed | Surveillance Center apps (1663 and 1661) list AS65; 4 free camera licences, more purchasable, per spec page. |
Calculated from the Lockerstor 12RD (AS6512RD)'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 | Q8 | Yes | 19GB | Not measured | Estimated |
| Qwen3 8BQwen, 8.19B | Q8 | Yes | 19GB | Not measured | Estimated |
| Gemma 3 12BGemma, 11.8B | Q8 | Yes | 24GB | Not measured | Estimated |
| Qwen3 14BQwen, 14.8B | Q8 | Yes | 28GB | Not measured | Estimated |
| Phi-4 14BPhi, 14.7B | Q8 | Yes | 28GB | Not measured | Estimated |
| Mistral Small 3 24BMistral, 23.6B | Q8 | Yes | 38GB | Not measured | Estimated |
| Gemma 3 27BGemma, 27B | Q8 | Yes | 44GB | Not measured | Estimated |
| Qwen3 30B A3BQwen, 30.5B | Q8 | Yes | 47GB | Not measured | Estimated |
| Qwen3 32BQwen, 32.8B | Q8 | Yes | 50GB | Not measured | Estimated |
| DeepSeek-R1 32B (Qwen distill)DeepSeek, 32.8B | Q8 | Yes | 50GB | Not measured | Estimated |
| Llama 3.3 70BLlama, 70.6B | Q8 | Yes | 98GB | Not measured | Estimated |
| Llama 4 Scout 17Bx16ELlama, 109B | Q4 | Yes | 89GB | Not measured | Estimated |
| Qwen2.5 72BQwen, 72.7B | Q8 | Yes | 101GB | Not measured | Estimated |
| Mistral Large 2 123BMistral, 123B | Q5 | Yes | 112GB | Not measured | Estimated |
One verdict per workload, from the Lockerstor 12RD (AS6512RD)'s use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
The Lockerstor 12RD is an acceptable local LLM host, with limits. Support is marked limited and vendor confirmed: Ollama and Open WebUI are listed for the AS65 series, but inference is CPU-only and ASUSTOR recommends an external GPU host.
Private RAG is workable here, not a strong match. ASUSTOR lists RAG as limited and vendor confirmed, with vector databases such as Chroma, pgvector and Weaviate running via Docker, and it recommends external compute for inference.
AI search is a poor fit on the Lockerstor 12RD. No AI or semantic file search feature was found in ADM 5.1 material, and that row is Unverified, so treat it as absent.
AI photos work acceptably on the Lockerstor 12RD. The Photo Gallery app lists the AS65 series, and ADM 5.1 adds facial recognition grouping powered by locally run AI, which is vendor confirmed.
Surveillance is an acceptable use for the Lockerstor 12RD. The Surveillance Center apps list the AS65 series with 4 free camera licences and more purchasable, vendor confirmed, and 12 bays hold recordings.
Home lab use is acceptable. Docker and virtual machines are both supported, the memory can be expanded to 128GB, and the 4 network ports allow separate traffic paths.
Business use is acceptable for file storage and services. It has 12 bays, two redundant 350W supplies and a 3 year warranty, though memory is non-ECC and networking tops out at 2.5GbE.
Creative AI is workable only at a modest level. GPU expansion is vendor confirmed as not supported and hardware acceleration is Unverified as absent, so work relies on CPU inference.
The Lockerstor 12RD (AS6512RD) ships with 8GB of DDR4-2133 SO-DIMM and supports up to 128GB. Upgrading to the maximum lifts the model ceiling from small models up to around 7B parameters to 70B class models at higher-quality quantization, or large mixture-of-experts models at 4-bit (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, 128GB leaves space for a model of roughly 200B parameters at most (Estimated).
No. The Lockerstor 12RD (AS6512RD) has no PCIe slot or OCuLink port for a graphics card. Inference runs on the CPU, so keep to small models.
With limits. The Lockerstor 12RD (AS6512RD) 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 + 2 x 1GbE. Fine for one or two users; indexing a large archive over the network will be slower.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| 8GB | 8GB | None | Diskless; 1x 8GB DDR4 pre-installed, 4 slots. |
The closest ASUSTOR model by bay count and AINAS Score. Only the specifications that differ are listed.
The Lockerstor 12RD (AS6512RD) scores best for Home Lab (54/100) and Business (50/100). It suits home lab owners who want one box for storage, containers, media and a few AI services.
It is a weaker fit for buyers who expect to add a graphics card later and for teams moving large files over the network every day.
Other AI Capable systems closest in price.
Buying guides that list it
Yes, on memory alone. Mistral Large 2 123B needs 73GB at Q4 and fits within the 128GB ceiling (Estimated). It does not fit at Q8, which needs 130GB (Estimated). It ships with 8GB, so you must upgrade first. This is a memory calculation, not a benchmark, and inference is CPU-only.
Yes. The memory is expandable to 128GB. It uses DDR4-2133 SO-DIMM, and the 8GB configuration has one 8GB module pre-installed with 4 slots in total. It is not ECC memory, according to the spec page.
No. ASUSTOR states its NAS models do not feature a discrete GPU. The single PCIe 3.0 x8 slot runs at an x4 link and has no documented GPU use. The slots are documented for network and SAS cards. It has no OCuLink, Thunderbolt or USB4 ports either.
No. Estimated: Qwen3 235B A22B is 142GB at Q4 and 250GB at Q8, both beyond what the 128GB memory ceiling can hold, so it does not fit at either level. The largest listed model that fits is Mistral Large 2 123B at Q4 (Estimated), and Llama 3.3 70B fits even at Q8 once the memory is raised to 128GB (Estimated). This is a memory calculation, not a benchmark, and inference would run on the Atom C3538's 4 CPU cores only.
Yes. VirtualBox is listed for the AS65 series on ADM 5, and Docker Engine is supported too. With the Atom C3538's 4 cores and memory expandable to 128GB, virtual machines are practical for light lab work.
3 years. That is the manufacturer warranty stated by ASUSTOR. The warranty covers the NAS itself, not the drives you add, because it comes diskless. It was released in 2021. It has two redundant 350W power supplies.
Yes. Ollama support is documented by the manufacturer. See the capability table above for the source.
Yes. Open WebUI support is documented by the manufacturer. See the capability table above for the source.
Yes. AI photos support is documented by the manufacturer. See the capability table above for the source.
Yes. Surveillance AI support is documented by the manufacturer. See the capability table above for the source.
No. Hardware acceleration support is not yet verified. See the capability table above for the source.
It has 2 x 2.5GbE + 2 x 1GbE built in, and the 1x PCIe 3.0 x8 slot (x4 link) 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.