4GB max RAM · 1 x M.2 · Not stated
AINAS Score 16
No verified price

Traditional NASASUSTOR
Why Traditional NAS: ARM Realtek RTD1619B with a fixed 2GB of RAM, no NPU and no expansion; ASUSTOR's Ollama app is not offered for the AS33 v2 series.
Updated
Can the ASUSTOR Drivestor 4 Pro Gen2 (AS3304T v2) run local AI?
The ASUSTOR Drivestor 4 Pro Gen2 (AS3304T v2) is a Traditional NAS built for file storage and backup rather than local AI. With Realtek RTD1619B and a 2GB memory ceiling, it can run light services but not a useful local language model.
We list it so you can compare it with AI-capable systems at a similar price.
The Drivestor 4 Pro Gen2 suits someone who wants a plain four-bay file server with a bit of light media and camera work, not a local AI box. It has four bays, a quad-core Realtek RTD1619B ARM CPU, one 2.5GbE port and ADM 5. ASUSTOR confirms Photo Gallery with locally run facial recognition and Surveillance Center for up to 25 IP cameras. Docker is available. A 90W power supply and a 3 year warranty complete the picture. It scores 17 on the AINAS Score, which reflects how little AI headroom it has. There is no M.2 slot and no PCIe slot, so all storage goes through the four drive bays.
The one thing that rules it out is memory. The Drivestor 4 Pro Gen2 has 2GB of DDR4 soldered onto the board, with no way to upgrade it. Nothing on it can run a local language model, and ASUSTOR's Ollama and Open WebUI app is not offered for this series. Anyone who wants local LLMs, private document search or RAG should look for a NAS with upgradeable memory of 16GB or more, an x86 CPU or an NPU, and either a PCIe slot or a documented AI app. Those buying only for storage and cameras can ignore this limit.
Amazon UK
2GB
£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 | No | Unverified | 2GB fixed RAM on ARM; ASUSTOR's Ollama + Open WebUI app support list excludes AS33 v2. No evidence of local LLM use. |
| Ollama | No | Unverified | Ollama + Open WebUI app is not offered for AS33 v2 (support list: AS52/53/54/63/64/65/66/67/68/71/72/81, FS67/FS68). Docker is available but no evidence found. |
| Open WebUI | No | Unverified | Not offered for AS33 v2 in App Central. |
| RAG | No | Unverified | No RAG app for this series; 2GB RAM. |
| AI search | No | Unverified | No AI or semantic search feature found in ADM 5.1 material. |
| GPU expansion | No | Vendor confirmed | No PCIe slot or M.2 on spec page. |
| Hardware acceleration | Limited | Vendor confirmed | ASUSTOR says the upgraded iGPU improves multimedia transcoding; no NPU. No AI use of the GPU documented. |
| AI photos | Yes | Vendor confirmed | Photo Gallery 4.2.0 (app 1254, ADM 5.1.1+) lists AS33v2; changelog mentions facial recognition, which ADM 5.1 describes as locally run AI. |
| AI video | No | Unverified | No AI video feature found in ADM documentation. |
| Surveillance AI | Yes | Vendor confirmed | Surveillance Center 2.8.4 (app 1715) and Surveillance Center 3 (app 1732) list AS33v2; up to 25 IP cameras per ASUSTOR licence table. |
Calculated from the Drivestor 4 Pro Gen2 (AS3304T v2)'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 | None fits | No | 11GB at Q4 | Not measured | Estimated |
| Llama 3.1 8BLlama, 8.03B | None fits | No | 14GB at Q4 | Not measured | Estimated |
| Qwen3 8BQwen, 8.19B | None fits | No | 15GB at Q4 | Not measured | Estimated |
| Gemma 3 12BGemma, 11.8B | None fits | No | 18GB at Q4 | Not measured | Estimated |
| Qwen3 14BQwen, 14.8B | None fits | No | 20GB at Q4 | Not measured | Estimated |
| Phi-4 14BPhi, 14.7B | None fits | No | 19GB at Q4 | 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 Drivestor 4 Pro Gen2 (AS3304T v2)'s use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
The Drivestor 4 Pro Gen2 is a poor choice for running local language models. Local LLM support is unconfirmed, the 2GB of fixed memory is far below what listed models need, and ASUSTOR's Ollama app excludes this series.
Private RAG is a poor fit here. ASUSTOR offers no RAG app for this series, Open WebUI is not offered in App Central, and 2GB of memory leaves no room for a model plus an index.
AI search is a poor fit on this system. No AI or semantic search feature turned up in the ADM 5.1 material, so that capability remains unconfirmed.
The verdict for AI photos is poor, even though a basic feature exists. Photo Gallery 4.2.0 lists this model and ASUSTOR confirms facial recognition, which ADM 5.1 describes as locally run, but nothing wider is documented.
Surveillance is rated poor overall, despite vendor-confirmed support. Surveillance Center lists this model with up to 25 IP cameras, but there is no NPU and no documented AI use of the GPU for detection.
Home lab use is poor. Docker is available, but memory is fixed at 2GB, virtual machines are not supported, and there is no M.2 slot or PCIe slot to expand into.
Business use is rated poor. The memory has no ECC, it is fixed at 2GB, and the single 2.5GbE port gives no network redundancy, though the 3 year warranty is a plus.
Creative AI is a poor fit. ASUSTOR confirms there is no GPU expansion, hardware acceleration is limited to media transcoding with no AI use documented, and local LLM support is unconfirmed.
The Drivestor 4 Pro Gen2 (AS3304T v2) ships with 2GB of DDR4 (onboard) that cannot be upgraded, so the configuration you buy is the one you keep. For local AI that matters more than anything else on the spec sheet: model size is capped by memory, not storage.
No. The Drivestor 4 Pro Gen2 (AS3304T v2) has no PCIe slot or OCuLink port for a graphics card. Inference runs on the CPU, with Integrated GPU (model not named by ASUSTOR) handling media tasks, so keep to small models.
Not well. A useful private RAG setup needs more memory or compute than the Drivestor 4 Pro Gen2 (AS3304T v2) offers.
Networking is 2.5GbE. Fine for one or two users; indexing a large archive over the network will be slower.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| 2GB | 2GB | None | Diskless; 2GB DDR4 onboard, not expandable. |
The closest ASUSTOR model by bay count and AINAS Score. Only the specifications that differ are listed.
The Drivestor 4 Pro Gen2 (AS3304T v2) scores best for Home Lab (21/100) and Business (21/100). 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 Traditional NAS systems closest in price.
No. Llama 3.2 3B is the smallest model listed on this page, and it needs 2GB at Q4 and 3.4GB at Q8. The Drivestor 4 Pro Gen2 has 2GB in total, so it does not fit at either level (Estimated). Nothing on the system can run model software anyway, because ASUSTOR's Ollama app is not offered for this series.
No. The 2GB of DDR4 is soldered onto the board, and the maximum is also 2GB. ASUSTOR's spec lists it as not expandable. Anyone who needs more memory for apps or containers has to choose a different NAS with a SODIMM slot or a higher fixed amount.
Yes. ASUSTOR lists Docker Engine for this model, and it needs ADM 5.1.0 or later. Memory is fixed at 2GB, so containers have little room. Virtual machines are not supported, because the VirtualBox apps list x86 series only. No Ollama app is offered for this series.
No. It has zero M.2 slots and no PCIe slot, so there is no way to add SSD caching or a storage pool on NVMe. All storage goes through the four drive bays. The spec page also lists no Thunderbolt, USB4 or OCuLink for external expansion.
No. The Drivestor 4 Pro Gen2 uses a quad-core Realtek RTD1619B ARM CPU with an integrated GPU that ASUSTOR does not name, and there is no NPU. ASUSTOR says the upgraded iGPU improves multimedia transcoding, and that is vendor confirmed, but no AI use of the GPU is documented. With a fixed 2GB of memory and no NPU, there is no hardware route to local language models, and ASUSTOR's Ollama app is not offered for this series either.
3 years. That is the warranty period listed by ASUSTOR for this model. The system was released in 2024 and is sold diskless with 2GB of onboard memory, so the warranty covers the NAS itself and not the drives you add.
No. Ollama support is not yet verified. See the capability table above for the source.
No. Open WebUI support is not yet verified. 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 2.5GbE built in. The fastest port is 2.5GbE.
Every field records its source. Specifications were last checked on . Spot an error? Send a correction with a manufacturer link.