64GB max RAM · 12 bays · 2 x 10GbE
AINAS Score 52
No verified price
AI ReadyiXsystemsDiscontinued
Why AI Ready: AMD EPYC 3rd Gen per a retailer (iXsystems gives core counts only, 16 to 64), PCIe Gen4 x16 slots with no vendor-documented GPU path, and a 1024GB memory ceiling meet the EPYC rule for AI Ready; not native because iXsystems documents no GPU inference.
Updated
Can the iXsystems TrueNAS R30 run local AI?
Partly. The iXsystems TrueNAS R30 runs local models on AMD EPYC 3rd Gen with 3D V-Cache per retailer, 16 to 64 cores alone, and its 1024GB ceiling allows 70B class models at higher-quality quantization, or large mixture-of-experts models at 4-bit (Estimated), but replies will be slow without graphics acceleration.
No graphics card can be fitted, which keeps every model on the CPU.
Local LLM support is not yet verified.
The TrueNAS R30 suits buyers who want an all-NVMe rackmount storage server that can also host container apps on the same box. It has 16 NVMe slots, an AMD EPYC CPU with up to 64 cores, and DDR4-3200 RDIMM memory expandable to 1024GB with ECC. That memory ceiling is the main AI draw. Every model listed on this page fits at Q4 and Q8 (Estimated), up to Qwen3 235B A22B. It runs TrueNAS with Docker and VM support, and Ollama, Open WebUI, Immich and Frigate are listed in the TrueNAS Apps catalog. The AINAS Score is 52.
The R30 has no GPU path that iXsystems documents, and that rules it out for anyone who wants accelerated inference. It has a PCIe 4.0 x16 full-height slot, a second x16 half-height slot and an x8 half-height slot, but GPU use is Unverified, and there is no NPU or integrated GPU. iXsystems has also discontinued the R30. Networking on the rear is only 2 x 1GbE, so moving large model files or serving many users needs a network card in one of those slots. Buyers who want fast local inference should look for a system with a documented GPU slot and stated card power, or an NPU with a published TOPS figure, plus 10GbE or faster networking on board. Buyers who mainly want a large memory pool for CPU work are less affected.
No verified price
We have not verified a listing for the iXsystems TrueNAS R30 at a retailer we track. The specifications below stand on their own; a price appears here as soon as a listing is confirmed.
Product 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 | Unverified | Docker apps on an EPYC CPU of up to 64 cores with 1024GB maximum; Ollama app exists in the TrueNAS catalog but no vendor statement of LLM use on this model. Enterprise licensed systems need apps enabled in the licence. |
| Ollama | Limited | Vendor confirmed | Ollama listed in the TrueNAS Apps catalog (community train, TrueNAS 24.10.2.2 or newer). Community support only. |
| Open WebUI | Limited | Vendor confirmed | Open WebUI listed in the TrueNAS Apps catalog (community train). Community support only. |
| RAG | Limited | Unverified | Possible via Open WebUI; no vendor or platform-specific evidence found. |
| AI search | No | Unverified | No AI or semantic search feature documented on the R-Series page. |
| GPU expansion | Limited | Unverified | PCIe 4.0 x16 full-height slot exists; GPU use is not documented by iXsystems. |
| Hardware acceleration | No | Unverified | No GPU, NPU or accelerator listed in R-Series specifications. |
| AI photos | Limited | Vendor confirmed | Immich listed in the TrueNAS Apps catalog (community train). |
| AI video | No | Unverified | No AI video feature documented. |
| Surveillance AI | Limited | Vendor confirmed | Frigate NVR listed in the TrueNAS Apps catalog (community train). No detection accelerator listed. |
Calculated from the TrueNAS R30'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 | Q8 | Yes | 149GB | Not measured | Estimated |
| Qwen2.5 72BQwen, 72.7B | Q8 | Yes | 101GB | Not measured | Estimated |
| Mistral Large 2 123BMistral, 123B | Q8 | Yes | 164GB | Not measured | Estimated |
One verdict per workload, from the TrueNAS R30's use-case scores and capability rows. Unknown means we cannot confirm it, not that it probably works.
The R30 is an acceptable host for a local LLM, but it is not confirmed for that job. The capability row is limited and Unverified, and the Ollama app is community supported only, though memory reaches 1024GB.
Private RAG is workable on the R30, with caveats. The RAG row is limited and Unverified, as it depends on Open WebUI, which is vendor listed in the Apps catalog with community support only.
The R30 is a poor choice for AI search. No AI or semantic search feature is documented on the R-Series page, and the capability row is marked no and Unverified.
AI photo management is acceptable on the R30. Immich is listed in the TrueNAS Apps catalog, vendor confirmed, but support is limited and community run, and Docker is available.
The R30 is acceptable for surveillance. Frigate NVR is vendor confirmed in the Apps catalog as limited support, and no detection accelerator is listed, so detection would run on the CPU.
The R30 is an acceptable home lab machine. It offers Docker and VM support, up to 64 cores, ECC RDIMM memory, and three PCIe 4.0 slots, though apps need to be enabled in the licence.
The R30 is acceptable for business storage. It combines 16 NVMe slots, ECC memory, an 800W power supply and TrueNAS Enterprise, but it is discontinued and offers only 2 x 1GbE on the rear.
We cannot confirm the R30 for creative AI. GPU expansion is limited and Unverified, as iXsystems documents no GPU use, and the hardware acceleration row says no.
The TrueNAS R30 can be bought without memory and takes up to 1024GB of DDR4-3200 RDIMM with ECC. Fitting the maximum is what turns it into a serious local AI machine.
Working backwards from 1024GB: reserve about 8GB for the operating system and storage services, and at roughly 0.6GB per billion parameters at 4-bit, the largest model that fits is around 1693B (Estimated).
ECC memory support is a plus for always-on systems that also hold important files.
There is no supported way to add a GPU to the TrueNAS R30: no PCIe slot and no OCuLink. Inference runs on the CPU, so keep to small models.
For a personal library, yes. A compact embedding model and a small chat model will run, but the TrueNAS R30 is better as the document store behind a RAG system whose model runs on stronger hardware.
Built-in networking: 2 x 1GbE. Adequate for a household, slower for bulk indexing.
| Configuration | RAM | SSD | Notes |
|---|---|---|---|
| Up to 1024GB, quote configured | 1024GB | None |
The closest iXsystems model by bay count and AINAS Score. Only the specifications that differ are listed.
Its strongest AINAS profiles are RAG at 58 and Home Lab at 56, which points to anyone building private search and question answering over their own documents.
Look elsewhere if you are buyers who expect to add a graphics card later, or teams moving large files over the network every day.
Other AI Ready systems closest in overall score.
Yes, on paper. Qwen3 235B A22B needs 142GB at Q4 and 250GB at Q8, and the R30 takes up to 1024GB of memory, so both fit (Estimated). This is a calculation, not a benchmark, and no speed is stated. The local LLM capability is limited and Unverified, so treat it as a memory fit only.
Yes. The R30 uses DDR4-3200 RDIMM across 16 slots and the memory is expandable up to 1024GB. A typical configuration is 256GB, but there is no fixed amount, and the starting capacity is set when the system is configured. The 1024GB ceiling is per controller, according to iXsystems.
Yes, it takes RDIMM memory, which is ECC. Confidence on this point is low, because iXsystems does not state ECC outright and the evidence comes from a third-party source. Confirm the ECC specification with iXsystems before relying on it.
Not confirmed. It has one PCIe 4.0 x16 full height slot, one x16 half height slot and one x8 half height slot, but iXsystems documents no GPU use. The gpu_expansion capability is limited and Unverified, and there is no OCuLink, Thunderbolt or USB4.
16. The R30 is an all-NVMe system with no conventional drive bays. iXsystems lists 16 NVMe slots, and a third-party source describes them as 12 hot-swap plus 4 internal. That gives fast flash storage for model files and apps, and the 1024GB memory ceiling leaves room to cache them. Networking is 2 x 1GbE on the rear, so that flash speed only reaches other machines in full through a faster network card in a PCIe slot.
Yes. The R30 runs TrueNAS with Docker apps and VM support through KVM. Apps run on Enterprise licensed systems only when enabled in the licence, and VM licensing is not confirmed. Ollama, Open WebUI, Immich and Frigate are in the Apps catalog on the community train.
Partly. Ollama support is documented by the manufacturer. See the capability table above for the source.
Partly. Open WebUI support is documented by the manufacturer. See the capability table above for the source.
Partly. AI photos support is documented by the manufacturer. See the capability table above for the source.
Partly. 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 1GbE built in, and the 1x PCIe 4.0 x16 full height, 1x PCIe 4.0 x16 half height, 1x PCIe 4.0 x8 half height, OCP 3.0 slot not used can take a faster network card. The fastest port is 1GbE.
Every field records its source. Specifications were last checked on . Spot an error? Send a correction with a manufacturer link.