64GB max RAM · 12 bays · 2 x 10GbE
AINAS Score 47
No verified price

AI CapableQNAP
Why AI Capable: Atom C5125 is not a current desktop-class CPU; QNAP's GPU list for its PCIe Gen3 x4 slots tops out at a 4GB RTX A400.
Updated
Can the QNAP TS-855eU run local AI?
Only lightly. Expect 32B class models comfortably, and 70B class models at 4-bit with little headroom (Estimated) on the QNAP TS-855eU: Intel Atom C5125 does the work with up to 64GB of memory, which is fine for experiments rather than a daily assistant.
Its PCIe x4 slot leaves room to add a GPU when you outgrow CPU inference.
Local LLM support is documented by the manufacturer.
The TS-855eU suits small offices and home labs that want an 8-bay 2U short-depth rack NAS with room to grow into local AI. It has an Intel Atom C5125 with 8 cores, 8GB of DDR4 SODIMM memory that expands to 64GB, and ECC is supported. It has two M.2 slots, two PCIe Gen3 x4 slots and 2x 2.5GbE. Docker and virtual machines are supported. Ollama and Open WebUI are vendor confirmed, and with 64GB Llama 3.3 70B fits at Q4 (Estimated). It carries a 3 year warranty and a 250W power supply, and its AINAS Score is 46.
The main catch is the GPU path. The PCIe slots only take the low-profile cards on QNAP's list, and the largest is a 4GB RTX A400. That means inference is mostly on the CPU, and the vendor says so. Buyers who want fast models in graphics memory, or who plan to run large models interactively, should look for a system with a full-length PCIe slot or integrated graphics, a much higher VRAM ceiling, or an NPU. The TS-855eU has no NPU and no integrated GPU. Anyone who needs 10GbE should also note it ships with 2.5GbE only, so faster networking needs an adapter.
Amazon US
TS-855eU-8G
Ships from the US; delivery and import charges shown at checkout
$1,699.00
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.
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 | QNAP Ollama guide (x86, optional NVIDIA GPU). 64 GB RAM but supported GPUs top out at 4 GB VRAM, so mostly CPU inference. |
| Ollama | Yes | Vendor confirmed | Container Station Ollama tutorial for x86_64 QTS/QuTS hero. |
| Open WebUI | Yes | Vendor confirmed | QNAP Open WebUI tutorial, QTS 5.2 / QuTS hero h5.2+. |
| RAG | Limited | Unverified | Qsirch AI Knowledge Search (RAG) uses cloud LLMs; not confirmed for this model. |
| AI search | Yes | Vendor confirmed | Qsirch AI Semantic Search needs 64-bit x86 and 8 GB RAM. |
| GPU expansion | Yes | Vendor confirmed | GPGPU list: GT 1030 LP, Quadro P620/P1000, RTX A400 4GB (low-profile, 2U). |
| Hardware acceleration | Limited | Vendor confirmed | No iGPU; Intel QAT; product page says optional Coral M.2 Accelerator for image recognition. |
| AI photos | Yes | Vendor confirmed | QuMagie AI-assisted photo management on both OSes. |
| AI video | Yes | Vendor confirmed | QVR Face, QVR Human and QVR DoorAccess listed under QTS. |
| Surveillance AI | Yes | Vendor confirmed | QVR Pro 8 free channels (QTS); QVR Surveillance 2 free + 6 on registration. |
Calculated from the TS-855eU'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 | Q4 | Yes | 60GB | Not measured | Estimated |
| Llama 4 Scout 17Bx16ELlama, 109B | None fits | No | 89GB at Q4 | Not measured | Estimated |
| Qwen2.5 72BQwen, 72.7B | Q3 | Yes | 54GB | Not measured | Estimated |
| Mistral Large 2 123BMistral, 123B | Q2 | Yes | 62GB | Not measured | Estimated |
One verdict per workload, from the TS-855eU's use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
The TS-855eU is an OK choice for local LLMs, with limits. QNAP's Ollama guide is vendor confirmed, but supported GPUs top out at 4GB of VRAM, so inference is mostly on the CPU.
Private RAG is workable here, but not fully confirmed. Open WebUI and Ollama are vendor confirmed, while Qsirch AI Knowledge Search uses cloud LLMs and is Unverified for this model.
AI search is a reasonable fit on the TS-855eU. Qsirch AI Semantic Search is vendor confirmed and needs 64-bit x86 and 8GB of RAM, which the included memory meets.
AI photo management is an OK use for this system. QuMagie AI-assisted photo management is vendor confirmed on both QuTS hero and QTS.
Surveillance is an OK fit. QVR Pro offers 8 free channels on QTS, while QVR Surveillance gives 2 free and 6 more on registration, and QVR Face and QVR Human are listed.
Home lab use is OK for the TS-855eU. It supports Docker and virtual machines, has 8 bays, 2 M.2 slots and 2 PCIe Gen3 x4 slots, and GPU pass-through is listed under QTS 6.1.0 only.
Business use is OK, particularly for compact sites. It is a 2U short-depth rack unit with ECC support, a 3 year warranty and a redundant power supply variant.
Creative AI is only an OK fit. GPU expansion is vendor confirmed, but the card list tops out at a 4GB RTX A400, and hardware acceleration is limited with no iGPU.
Plan on upgrading. 8GB is fitted as standard, while the TS-855eU accepts 64GB of DDR4 SODIMM, ECC included, which is the difference between small models up to around 7B parameters and 32B class models comfortably, and 70B class models at 4-bit with little headroom (Estimated).
Our estimate for the biggest model that fits in 64GB is about 93B parameters at 4-bit, after setting aside 8GB for the system. Long context windows reduce that figure (Estimated).
ECC support protects against silent memory errors on a box that runs around the clock.
Only a small one. The 2x PCIe Gen3 x4 in the TS-855eU is sized for network and storage cards; low-profile GPUs are the limit.
As the storage half of a RAG setup it works well; as the model host it is limited. Small embedding models run fine on the TS-855eU, while the answering model needs more compute than it has.
The manufacturer also provides AI search across your files (documented by the manufacturer), which covers simple document questions without setting up your own stack.
For network access it has 2 x 2.5GbE. Enough for light use; large imports will take a while.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| TS-855eU-8G | 8GB | None | Single 250 W PSU. |
| TS-855eU-RP-8G | 8GB | None | Redundant PSU variant; compatibility index lists TS-855eU-RP separately. |
The closest QNAP model by bay count and AINAS Score. Only the specifications that differ are listed.
Best fit: home lab owners who want one box for storage, containers, media and a few AI services. That is where the TS-855eU scores highest (Home Lab 52/100), followed by Business (50/100).
Less suited to teams moving large files over the network every day.
Other AI Capable systems closest in price.
Buying guides that list it
Yes at Q4, as an estimate. Llama 3.3 70B needs 43GB at Q4 and the TS-855eU takes up to 64GB, so it fits (Estimated). At Q8 it needs 75GB and does not fit (Estimated). Llama 4 Scout 17Bx16E does not fit at Q4 (Estimated). With only 4GB GPUs supported, inference is mostly on the CPU.
64GB is the maximum. The TS-855eU ships with 8GB of DDR4 SODIMM and has 2 slots, so the 64GB ceiling is 2x 32GB. The memory sits in slots rather than being soldered. The compatibility list modules are DDR4-3200.
Yes, ECC is supported. The TS-855eU ships with 8GB of non-ECC memory, so you would need to fit compatible ECC modules to use it. Memory is in 2 DDR4 SODIMM slots, with a maximum of 64GB.
Yes, but only low-profile cards. It has 2 PCIe Gen3 x4 slots and the vendor GPGPU list is GT 1030 LP, Quadro P620, Quadro P1000 and RTX A400 4GB. It has no integrated GPU, and HDMI output needs an optional PCIe graphics card.
Two. The TS-855eU-8G has a single 250W power supply, and the TS-855eU-RP-8G adds redundant power supplies, and QNAP's compatibility index treats the RP model as a separate entry. Both come with 8GB of DDR4 SODIMM and take up to 64GB across 2 slots. The rest of the specification is shared: 8 bays, an 8-core Intel Atom C5125, 2 M.2 slots, two PCIe Gen3 x4 slots and 2x 2.5GbE. QNAP does not state the wattage of the RP power supplies, so check the spec sheet if your rack has a power budget.
Yes, 2. They are M.2 2280 PCIe Gen3 x4 slots. They sit alongside the 8 hot-swap 3.5in SATA bays. The two PCIe Gen3 x4 slots are separate, and are for add-in cards such as network adapters or GPUs.
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.
Partly. Hardware acceleration support is documented by the manufacturer. See the capability table above for the source.
It has 2 x 2.5GbE built in, and the 2x PCIe Gen3 x4 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.