32GB max RAM · 2 bays · 2 x 2.5GbE
AINAS Score 33
Price awaiting re-verification
Last checked

AI CapableASUSTOR
Why AI Capable: Intel Celeron N5105 (4C/4T) with Intel UHD Graphics, no NPU, no PCIe or GPU expansion and a 16GB memory ceiling; small CPU-only models via ASUSTOR's Ollama app at best.
Updated
Can the ASUSTOR Nimbustor 2 Gen2 (AS5402T) run local AI?
Only lightly. Expect 7B to 8B class models (Estimated) on the ASUSTOR Nimbustor 2 Gen2 (AS5402T): Intel Celeron N5105 does the work with up to 16GB of memory, which is fine for experiments rather than a daily assistant.
Without any GPU option, the processor carries the whole inference load.
Local LLM support is documented by the manufacturer.
The Nimbustor 2 Gen2 suits buyers who want a compact two-bay NAS with fast networking and light, CPU-only AI as an extra. It has an Intel Celeron N5105 with 4 cores, two 2.5GbE ports, four M.2 slots, Docker and virtual machine support on ADM 5. ASUSTOR's Ollama and Open WebUI app is vendor confirmed for this model and runs on the CPU cores alone. Photo Gallery face grouping and Surveillance Center are also vendor confirmed, and the licence table allows up to 44 IP cameras. Memory is upgradable SO-DIMM, from the 4GB it ships with up to 16GB. Small models are the realistic target: Qwen3 8B fits at Q4 (Estimated) once memory is upgraded. A 3 year warranty applies. The AINAS Score is 33. It draws power from a 65W supply.
The memory ceiling is the one thing that rules the Nimbustor 2 Gen2 out for some buyers. It affects anyone who wants larger models, because the ceiling is 16GB and memory cannot go beyond it. Gemma 3 12B is only marginal at Q4 (Estimated), and Mistral Small 3 24B does not fit at Q4 (Estimated). There is no PCIe slot, no discrete GPU option and no NPU, so a graphics card cannot make up the gap. ASUSTOR itself also advises running inference on a separate machine. Buyers who need larger models should look for a higher memory ceiling, a PCIe slot or an OCuLink or Thunderbolt port for a graphics card, or an NPU. Anyone happy with small models and file serving can ignore this.
Amazon UK
4GB
£332£369
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 | The Nimbustor 2 Gen2 (AS5402T) is on App Central's AS54 list for the Ollama and Open WebUI app, which runs on the N5105 cores alone. Its 4GB starting memory needs upgrading before any model loads, and the 16GB official limit caps model size. No model is preinstalled. |
| Ollama | Yes | Vendor confirmed | AS54 appears among the series supported by ASUSTOR's Ollama and Open WebUI package, which sets up Ollama as a Docker container on this two-bay unit. |
| Open WebUI | Yes | Vendor confirmed | Open WebUI arrives in the same App Central package as Ollama, so there is no separate install on the AS5402T. |
| RAG | Limited | Vendor confirmed | ASUSTOR pitches its NAS line as storage for RAG knowledge bases, with vector databases such as Chroma, pgvector and Weaviate in Docker, and Open WebUI has document chat built in. ASUSTOR itself advises running inference on a separate machine. |
| AI search | No | Unverified | ADM 5.1 material for the Nimbustor 2 Gen2 describes no AI or semantic search across files. |
| GPU expansion | No | Vendor confirmed | The AS5402T spec page lists no PCIe slot, and ASUSTOR says its NAS models carry no discrete GPU, so graphics cards are not an option. |
| Hardware acceleration | Limited | Vendor confirmed | The N5105's Intel UHD Graphics handles 4K H.265 decoding and transcoding according to ASUSTOR. There is no NPU, and ASUSTOR does not document the iGPU for AI inference. |
| AI photos | Yes | Vendor confirmed | Photo Gallery 4.2 on ADM 5.1.1 or later supports the AS54 series, and ADM 5.1 groups faces using AI that runs on the NAS itself. |
| AI video | No | Unverified | ASUSTOR's ADM documentation lists no AI video analysis for this model. |
| Surveillance AI | Yes | Vendor confirmed | Surveillance Center 2.8.4 (app 1663) and Surveillance Center 3 (app 1661) both list AS54. The spec page licence table allows up to 44 IP cameras. |
Calculated from the Nimbustor 2 Gen2 (AS5402T)'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 Nimbustor 2 Gen2 (AS5402T)'s use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
Poor for local LLM work. The Ollama and Open WebUI app is vendor confirmed but runs on the CPU cores alone, local LLM support is limited, and the 4GB starting memory needs upgrading before any model loads.
Poor for private RAG. Open WebUI document chat and Chroma in Docker are described by ASUSTOR, but RAG support is limited, and ASUSTOR itself advises running inference on a separate machine, and there is no NPU to help.
Poor for AI search, with no feature to point to. ADM 5.1 material describes no AI or semantic search across files, and the capability row is marked Unverified, so it is unconfirmed rather than a tested gap.
Poor for AI photos, despite a supported app. Photo Gallery 4.2 on ADM 5.1.1 or later is vendor confirmed and groups faces on the NAS, but that work falls to the four N5105 cores with no NPU.
Poor for surveillance, even though the apps are supported. Surveillance Center 2.8.4 and 3 are vendor confirmed and the licence table allows up to 44 IP cameras, but AI video analysis is Unverified and listed as unsupported.
Poor for home lab use in AI terms. Docker and virtual machines are both supported on ADM 5, but the 4 cores of the N5105 and the 16GB memory ceiling leave little room for running several workloads together.
Poor for business use as an AI platform. There is no ECC memory, only 4GB comes installed, and the 3 year warranty is the main assurance, with no local AI capability beyond only small CPU-only models.
Poor for creative AI work. There is no PCIe slot and graphics cards are not an option, and the Intel UHD Graphics is documented by ASUSTOR only for 4K H.265 decoding and transcoding, not AI inference.
Plan on upgrading. 4GB is fitted as standard, while the Nimbustor 2 Gen2 (AS5402T) accepts 16GB of DDR4-2933 SO-DIMM, which is the difference between only very small models of around 1B to 3B parameters and 7B to 8B class models (Estimated).
Our estimate for the biggest model that fits in 16GB is about 13B parameters at 4-bit, after setting aside 8GB for the system. Long context windows reduce that figure (Estimated).
A graphics card is not an option on the Nimbustor 2 Gen2 (AS5402T), which has neither a PCIe slot nor OCuLink. Inference runs on the CPU, with Intel UHD Graphics handling media tasks, so keep to small models.
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 Nimbustor 2 Gen2 (AS5402T), while the answering model needs more compute than it has.
For network access it has 2 x 2.5GbE. Enough for light use; large imports will take a while.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| 4GB | 4GB | None | Diskless; 1x 4GB DDR4 pre-installed, 2 SO-DIMM slots. |
The closest ASUSTOR model by bay count and AINAS Score. Only the specifications that differ are listed.
Best fit: small offices that need shared storage with private AI tools and room to grow. That is where the Nimbustor 2 Gen2 (AS5402T) scores highest (Business 36/100), followed by Home Lab (34/100).
Less suited to anyone who wants to run models larger than about 8B parameters and buyers who expect to add a graphics card later.
Other AI Capable systems closest in price.
Yes, Qwen3 8B fits at Q4 (Estimated), needing 5.2GB. At Q8 it is marginal (Estimated), needing 8.9GB, so it loads with little headroom. The unit ships with 4GB, so memory must be upgraded first. ASUSTOR's Ollama and Open WebUI app runs on the N5105 cores alone, with no preinstalled model. This is a calculation, not a benchmark, and no speed is implied.
Yes. It uses DDR4-2933 SO-DIMM memory in two slots, with one 4GB module fitted. The upper limit is 16GB, listed as 2x 8GB. There is no ECC support. Upgrading is the first step before running any model through the Ollama app. ASUSTOR's own notes say the 4GB starting memory needs upgrading before any model loads, and the 16GB limit caps model size.
No. The spec page lists no PCIe slot, and ASUSTOR says its NAS models carry no discrete GPU. There is also no OCuLink, Thunderbolt or USB4 port to connect an external one. The only graphics is the integrated Intel UHD Graphics, which ASUSTOR documents for 4K H.265 decoding and transcoding rather than AI inference. Buyers who need GPU acceleration should look elsewhere.
Yes, four. The Nimbustor 2 Gen2 pairs its two drive bays with four M.2 NVMe slots, so flash storage can sit alongside the hard drives without taking a bay. There is no PCIe slot, so the M.2 slots are the only internal expansion. The unit comes diskless with 4GB of memory, which means drives and any memory upgrade are bought separately. Networking is two 2.5GbE ports, and the warranty is 3 years.
Yes. Docker Engine 29.6.2 for this series requires ADM 5.1.0 or later. VirtualBox is listed for it too, so virtual machines are supported on ADM 5. ASUSTOR's Ollama and Open WebUI package sets up Ollama as a Docker container. With a 4 core Celeron N5105 and a 16GB memory ceiling, containers and virtual machines share limited resources.
No. It has no NPU, and ASUSTOR lists none. AI work runs on the four cores of the Intel Celeron N5105, with Intel UHD Graphics handling 4K H.265 decoding and transcoding. ASUSTOR does not document the integrated graphics for AI inference. Photo Gallery face grouping runs on the NAS itself without an NPU, and Ollama runs on the CPU cores alone.
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. 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.