128GB max RAM · 4 bays · 2 x 2.5GbE
AINAS Score 40
No verified price

AI CapableSynology
Why AI Capable: Intel Xeon D-1531 (2015 Broadwell-DE, 6 cores, no NPU, no iGPU) with a 48 GB memory ceiling (below 64 GB) and a PCIe Gen3 x8 slot that Synology supports only for NICs, M.2 and FC adapters, so no GPU path; CPU-only local AI via Docker.
Updated
Can the Synology DS3622xs+ run local AI?
The Synology DS3622xs+ can run small local models, with limits. Its Intel Xeon D-1531 and 48GB memory ceiling suit models up to the 14B class, with 32B class models a tight fit (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 confirmed by community reports.
The DS3622xs+ suits buyers who want a 12-bay Synology for storage first, with light CPU-only AI on the side. It has an Intel Xeon D-1531 with 6 cores, 16GB of DDR4 ECC memory that expands to 48GB, and 2 x 10GbE plus 2 x 1GbE. A PCIe Gen3 x8 slot takes extra network cards or M.2 adapters. DSM runs Docker and virtual machines, and Surveillance Station is listed for up to 90 IP cameras. Synology Photos recognition runs on the CPU. Models up to Qwen3 32B fit at Q4 (Estimated). The warranty is 5 years. The AINAS Score is 40.
The limit is the missing GPU path. The DS3622xs+ has no integrated graphics and no NPU, and Synology lists the PCIe slot only for network cards, M.2 adapters and Fibre Channel adapters, so there is no documented graphics card support. Anyone planning fast local LLM work, creative AI or AI search that needs GPU models will be disappointed. Those buyers should look for a system with a supported GPU slot or integrated graphics, an NPU, and a memory ceiling of 64GB or more. Its 4 memory slots also arrive half filled with 2 x 8GB, so reaching 48GB means adding two 16GB modules.
Amazon UK
DS3622xs+ 16GB ECC
No current price
Checked , no price yet
Out of stock
Buy on AmazonAmazon US
DS3622xs+ 16GB ECC
Ships from the US; delivery and import charges shown at checkout
$3,449.99
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 | Community confirmed | CPU-only Ollama via Container Manager on DSM (guide is for x86 Plus models with Container Manager); 16 GB base RAM allows small to 7B quantised models slowly. No GPU. |
| Ollama | Yes | Community confirmed | Not in Package Center; ollama/ollama container runs in Container Manager on DSM (CPU only). |
| Open WebUI | Yes | Community confirmed | Open WebUI container documented on Synology DSM pointing at Ollama. |
| RAG | Limited | Unverified | No Synology local RAG on this model; AI Console bridges to cloud or OpenAI-compatible APIs. Open WebUI document chat via Docker plausible but not verified on this unit. |
| AI search | No | Vendor confirmed | Synology AI Search in Drive needs GPU-powered models, available only on supported systems with a GPU (AI Console 2.0 beta lists FS3420, FS6420, RS11626xs+, RS3626xs, RS4826xs+, RS6426xs+); this model is not listed. |
| GPU expansion | No | Vendor confirmed | PCIe Gen3 x8 (x8 link) slot exists but Synology lists only NICs, M.2 adapters and FC HBAs; no GPU support documented. |
| Hardware acceleration | No | Vendor confirmed | Xeon D-1531 has no processor graphics per Intel ARK and no NPU; no hardware transcoding or AI acceleration. |
| AI photos | Yes | Vendor confirmed | Synology Photos facial and object recognition listed (on-device, CPU). |
| AI video | Limited | Community confirmed | No on-NAS deep video analytics; Surveillance Station can use on-camera AI from Synology BC500/TC500 cameras. |
| Surveillance AI | Yes | Vendor confirmed | Surveillance Station up to 90 IP cameras (1080p H.264); 2 free licences. |
Calculated from the DS3622xs+'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 | Q4 | Yes | 32GB | Not measured | Estimated |
| DeepSeek-R1 32B (Qwen distill)DeepSeek, 32.8B | Q4 | Yes | 32GB | Not measured | Estimated |
| Llama 3.3 70BLlama, 70.6B | Q2 | Yes | 40GB | Not measured | Estimated |
| Llama 4 Scout 17Bx16ELlama, 109B | None fits | No | 89GB at Q4 | Not measured | Estimated |
| Qwen2.5 72BQwen, 72.7B | Q2 | Yes | 44GB | Not measured | Estimated |
| Mistral Large 2 123BMistral, 123B | None fits | No | 96GB at Q4 | Not measured | Estimated |
One verdict per workload, from the DS3622xs+'s use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
The DS3622xs+ is a poor choice for local LLM work. Support is limited and community confirmed only: Ollama runs CPU-only in Container Manager, with no GPU, and the 16GB base memory suits small to 7B quantised models slowly.
Private RAG is workable on the DS3622xs+, with caveats. Synology offers no local RAG on this model, and Open WebUI document chat via Docker is unverified here, though Open WebUI itself is community confirmed with Ollama.
AI search is a poor fit for the DS3622xs+. Synology AI Search in Drive needs GPU-powered models, the vendor confirms this model is not on the supported list, and the system has no GPU.
AI photos are a reasonable use of the DS3622xs+. Synology lists Photos facial and object recognition as supported, vendor confirmed, running on the CPU with no hardware acceleration.
Surveillance is a reasonable use of the DS3622xs+. Surveillance Station is vendor confirmed for up to 90 IP cameras at 1080p H.264 with 2 free licences, though deep video analytics are not run on the NAS.
The DS3622xs+ is an acceptable home lab system. DSM runs Docker and Virtual Machine Manager, memory is ECC and expands to 48GB, and 2 x 10GbE ports help with networked storage.
Business use is acceptable on the DS3622xs+. It has 12 bays, ECC memory, 2 x 10GbE, a 550W power supply and a 5 year warranty, which suit file serving and virtualisation.
Creative AI is a poor fit for the DS3622xs+. Synology documents no GPU support in the PCIe slot, and the Xeon D-1531 has no graphics or AI acceleration, so there is no hardware path.
The DS3622xs+ ships with 16GB of DDR4 ECC SODIMM and supports up to 48GB with ECC. Upgrading to the maximum lifts the model ceiling from 7B to 8B class models to models up to the 14B class, with 32B class models a tight fit (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, 48GB leaves space for a model of roughly 66B parameters at most (Estimated).
ECC support protects against silent memory errors on a box that runs around the clock.
No. The DS3622xs+ has no PCIe slot or OCuLink port for a graphics card. Inference runs on the CPU, so keep to small models.
With limits. The DS3622xs+ 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 10GbE + 2 x 1GbE. That is ample for indexing large libraries and for several people querying at once.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| DS3622xs+ 16GB ECC | 16GB | None | 16 GB DDR4 ECC SODIMM (2x 8 GB), 4 slots, max 48 GB. |
The closest Synology model by bay count and AINAS Score. Only the specifications that differ are listed.
The DS3622xs+ scores best for Home Lab (55/100) and Business (51/100). 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.
Other AI Capable systems closest in price.
Yes, at Q4 only. DeepSeek-R1 32B (Qwen distill) needs 20GB at Q4, which fits within the 48GB ceiling (Estimated). At Q8 it needs 35GB and is marginal, with little headroom (Estimated). It would run CPU-only through Container Manager, as there is no GPU. Llama 3.3 70B does not fit at either level (Estimated).
Yes. It ships with 16GB of DDR4 ECC SODIMM as 2 x 8GB, and has 4 slots. The maximum is 48GB, made up of 2 x 8GB and 2 x 16GB. The memory sits in slots rather than being soldered, so you can add modules after purchase.
Yes. The DS3622xs+ uses DDR4 ECC SODIMM, and the 16GB configuration is sold with ECC memory fitted. ECC helps protect stored data and virtual machines from memory errors, which matters for a 12-bay storage system.
No. It has one PCIe Gen3 x8 slot, but Synology lists only network cards, M.2 adapters and Fibre Channel adapters for it. The vendor confirms no GPU support is documented. There is no OCuLink, Thunderbolt or USB4 port to attach one either.
No. There are 0 onboard M.2 slots. NVMe cache needs an add-in card such as the M2D20 or E10M20-T1 in the PCIe slot. The E10M20-T1 combines M.2 with 10GbE. Using the slot for cache means it cannot hold another card.
Yes. Container Manager runs Docker on Synology DSM, and virtual machines are supported too. Ollama is not in Package Center, but the ollama/ollama container runs CPU only in Container Manager, which is community confirmed, and Open WebUI is documented pointing at it. Community notes say the 16GB fitted allows small models up to 7B quantised, running slowly on the 6-core Xeon D-1531. Raising memory towards 48GB widens the Estimated fits, up to Qwen3 32B at Q4.
Yes. Ollama support is confirmed by community reports. See the capability table above for the source.
Yes. Open WebUI support is confirmed by community reports. 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 documented by the manufacturer. See the capability table above for the source.
It has 2 x 10GbE + 2 x 1GbE built in, and the 1x PCIe Gen3 x8 slot (x8 link); Synology-listed cards: E10G30-F2/E10G30-T2/E10G18-T1 (10GbE), E25G30-F2 (25GbE), E10M20-T1, M2D20, compatible FC HBAs can take a faster network card. The fastest port is 2 x 10GbE.
Every field records its source. Specifications were last checked on . Spot an error? Send a correction with a manufacturer link.