Products · Zettlab · NAS
Zettlab D8 Ultra
- Processor
- Intel Core Ultra 5 125H
- Memory (RAM)
- 16, 32 or 96 GB
- Launch price
- $1,600
- Released
- 2026
Product page (opens in a new tab)Discontinued Full AI report
Configurations
Pick one to see which models it runs. More memory lets larger models fit; bandwidth sets how fast they answer.
| Configuration | Memory | Usable for AI | Bandwidth | LLMs that run well | Launch price |
|---|---|---|---|---|---|
| Intel Core Ultra 5 125H | 16.0 GB | 12.0 GB | 90 GB/s | 316 of 622 | $1,600 |
| Intel Core Ultra 5 125H | 32.0 GB | 26.2 GB | 90 GB/s | 350 of 622 | — |
| Intel Core Ultra 5 125H · upgraded to the maker's maximum | 96.0 GB | 83.5 GB | 90 GB/s | 375 of 622 | — |
Specifications 16.0 GB configuration
| Maker | Zettlab |
|---|---|
| Type | NAS |
| Processor | Intel Core Ultra 5 125H NAS (16 GB) |
| Memory (RAM) | 16.0 GB |
| Usable for AI models | 12.0 GB |
| Memory bandwidth | 90 GB/s |
| FP16 compute | 0.7 TFLOPS |
| Runs models on | CPU cores |
| Drive bays | 8x SATA (3.5/2.5 inch) + 2x M.2 NVMe (PCIe 4.0 x4; 2280/2242/2230) |
| NAS system | ZettOS |
| Containers (Docker) | Yes |
| Virtual machines | Yes |
| Memory upgrade | DDR5 SO-DIMM, up to 5600 MT/s (non-ECC) · 2 slots · up to 96.0 GB per the maker · non-ECC |
| Graphics card options | PCIe: 1x SFF-8654 connector on the rear, PCIe 4.0 x8 (external cabled port, not an internal card slot; needs an…. External GPU: Through the SFF-8654 PCIe 4.0 x8 port with an external dock and power supply. 2x USB4 are also fitted, but…. Maker-supported graphics cards: Maker wiki FAQ (archived 2026-01-02): 'We'll support GTX 40/50 series GPUs' (sic, i.e. NVIDIA 40/50 series; RTX A-series untested). No model list published. |
| Availability | Discontinued (sold used or refurbished) |
| Notes | Intel Core Ultra 5 125H. Maker lists 'Local LLM and AI' and offline AI Q&A for the Ultra models, and TechRadar's review ran Gemma, Phi, QwQ, DeepSeek-R1 and Llama models on the D6 Ultra itself; Core Ultra 5 125H with up to 96 GB DDR5, plus Docker and VMs for other runners. |
| Product page | zettlab.com (opens in a new tab)Discontinued |
Running AI on the Zettlab D8 Ultra
Yes, it runs AI models on the NAS itself. Models run on its CPU cores, and this configuration leaves about 12.0 GB for them after the NAS system and its storage services. The grades below assume the NAS isn’t busy with backups or media work at the same time. More memory is the upgrade that matters most: see the configurations.
Apps you can install to run models
| App | How it installs |
|---|---|
| AnythingLLM | From the NAS app store |
| Any Docker image | In a container (Docker) |
| Ollama / Open WebUI or other model runners as Docker containers | In a container (Docker) |
AI features from Zettlab
| Feature | What it does | Where the AI runs |
|---|---|---|
| ZettAI AI Search | Natural-language search across images, audio, video and documents stored on the NAS; maker says processing is 100% on-device. | On the NAS |
| ZettAI AI Categories | Automatic sorting and tagging of files and media. | On the NAS |
| ZettAI AI Transcription / AI Clip | Audio and video transcription and clip finding. | On the NAS |
| ZettAI AI Q&A (Knowledge QA) | Chat / question answering over the user's own files (RAG) with a built-in large language model that works offline. | On the NAS |
| Local LLM and AI | Maker's compare table lists 'Local LLM and AI' for the Ultra models only, and the headline feature list says 'Local LLM'; model names and speeds are not published. | On the NAS |
| ZettAI local LLM chat (model library) | Chat with a language model hosted on the NAS. Maker's compare table lists 'Local LLM and AI' for the Ultra models only. | On the NAS |
Best LLMs for the Zettlab D8 Ultra 16.0 GB
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Mellum2.1 Thinking (12B-A2.5B) | Q4_K_M | 9.1 GB | 28.2 tok/s | Runs well |
| 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Maple Preview (20B-A1B) | Q3_K_M | 11.6 GB | 55.4 tok/s | Runs welltight fit |
| 58Good quality (B) · 58 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | Ling 3.0 Tiny (8B-A1.3B) | Q5_K_M | 7.4 GB | 42.2 tok/s | Runs great |
| 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | Nemotron 3.5 Bolt E4B | Q4_K_M | 6.6 GB | 17.3 tok/s | Runs well |
| 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | LFM2.5 24B-A2B | Q2_K | 11.3 GB | 44.9 tok/s | Runs welltight fit |
| 67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | K2 Horizon 7B | Q3_K_M | 7.2 GB | 11.2 tok/s | Runs well |
| 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Gemma 4 E4B | Q4_K_M | 6.6 GB | 17.3 tok/s | Runs well |
| 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Kimi-VL A3B Thinking 2506 | Q4_K_M | 11.8 GB | 23.9 tok/s | Runs welltight fit |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | Ling Mini 2.0 (16B-A1.4B) | Q4_K_M | 11.8 GB | 46.9 tok/s | Runs welltight fit |
| 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE 4.5 21B-A3B Thinking | Q2_K | 10.8 GB | 37.6 tok/s | Runs welltight fit |
Best Image models for the Zettlab D8 Ultra 16.0 GB
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 76Great quality (A) · 76 of 100: how capable the model is at FP8, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.2 [klein] 4B | FP8 | 10.2 GB | 58 s | Slow |
| 84Great quality (A) · 84 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | FLUX 3 Image Klein | Q4 GGUF | 10.2 GB | 7.3 min | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | Z-Image Turbo | Q4 GGUF | 10.1 GB | 3.2 min | Very slow |
| 79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | ERNIE-Image Turbo | Q4 GGUF | 11.3 GB | 4.1 min | Very slowtight fit |
| 76Great quality (A) · 76 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | Z-Image | Q4 GGUF | 10.1 GB | 21.9 min | Very slow |
| 78Great quality (A) · 78 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | FLUX.2 [klein] 9B | Q4 GGUF | 8.0 GB | 2.4 min | Very slowencoder unloaded after use |
Best Video models for the Zettlab D8 Ultra 16.0 GB
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 49Fair quality (C) · 49 of 100: how capable the model is at FP8, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | LTX-Video 2B | FP8 | 10.8 GB | 24.2 min | Slowtight fit |
| 83Great quality (A) · 83 of 100: how capable the model is at FP8, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | Bernini-R A14B (video generation + editing) | FP8 | 10.7 GB | 15.1 h | Very slowtight fit · CPU offload |
| 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | HY-OmniWeaving (8.3B, unified video) | FP8 | 7.8 GB | 11.1 h | Very slowCPU offload |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 T2V 1.3B | FP16 | 8.4 GB | 4.5 h | Very slowencoder unloaded after use |
| 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | Wan 2.1 VACE 1.3B | FP16 | 10.1 GB | 4.9 h | Very slowencoder unloaded after use |
Best Decision models for the Zettlab D8 Ultra 16.0 GB
| Quality | Model | Quant | Memory | Speed | Runs |
|---|---|---|---|---|---|
| 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | OpenAI Privacy Filter | F16 | 3.8 GB | 49 ms | Runs great |
| 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | Llama Prompt Guard 2 22M | Q8_0 | 0.7 GB | 115 ms | Runs great |
| 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | Jina Reranker v3.5 | Q8_0 | 2.4 GB | 880 ms | Slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | GLiNER2.5 Decide | Q8_0 | 1.2 GB | 386 ms | Runs well |
| 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Jina Reranker v3 | Q8_0 | 2.4 GB | 880 ms | Slow |
| 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | Qwen3 Reranker 4B | Q8_0 | 6.1 GB | 5.9 s | Very slow |
| 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | FRIDA-Decisions | Q8_0 | 1.7 GB | 643 ms | Slow |
| 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | GLiClass Base v3 | Q8_0 | 0.8 GB | 152 ms | Runs well |
Other products with the Intel Core Ultra 5 125H NAS (16 GB)
Spec sources
Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.