HP ZBook Fury 15 G8 Review: 8 GB RTX A4000 Workstation for Sustained AI
Who is this laptop for?
The HP ZBook Fury 15 G8 is the sustained-workload 8 GB machine in our line-up. It pairs the RTX A4000 Laptop GPU (8 GB GDDR6) with an Intel Core i7-11850H (or Xeon W) in a true mobile-workstation chassis built for hours of continuous compute. Where the portable ASUS ROG Zephyrus G14 throttles on long runs, the Fury holds its clocks — making it the better choice when AI work means batch jobs and training, not bursts.
Students (Budget: £700–£1,000)
For most students this is more workstation than coursework needs, and it’s heavy at ~2.5 kg. But for a research student running long fine-tuning experiments or batch diffusion overnight, the Fury’s sustained thermals and ISV-grade reliability matter more than portability. If you only need bursts of GPU work on the move, the lighter G14 is the smarter spend.
ML Engineers & Data Scientists
This is a serious portable inference and light-training box. The RTX A4000 Laptop is a professional 8 GB Ampere GPU with ECC-capable memory and the same CUDA stack as the desktop A-series. SDXL, ComfyUI, quantised FLUX.1 and 13B LLMs in Q4 all run, and the workstation cooler sustains GPU clocks far longer than a thin gaming laptop. With 32 GB RAM (expandable to 128 GB across 4 slots) you can stage large datasets and run 30B-class models on CPU alongside GPU work.
Small Teams & Startups
As a shared, dependable AI workstation the Fury earns its keep. ISV certification, a robust service network, and HP’s enterprise warranty make it a safer team purchase than a consumer gaming laptop. It replaces a desktop for inference and light fine-tuning at a fraction of new-workstation cost. The trade-off is weight and a heavy 200W charger — this is a transportable workstation, not a daily carry.
What can it actually run?
| Task | Works? | Notes |
|---|---|---|
| GitHub Copilot / Cursor AI | ✅ Yes | API-based, runs perfectly |
| Whisper transcription (local) | ✅ Yes | large-v3 at ~1.3× realtime (GPU-accelerated) |
| Ollama 7B | ✅ Yes | Fully GPU-resident. ~42–52 tok/s (estimated) |
| Ollama 13B | ✅ Yes | Q4 fits in 8 GB at moderate context. ~17–23 tok/s (estimated) |
| Stable Diffusion XL | ✅ Yes | ~12–18s per 1024×1024 image; sustains across batches |
| ComfyUI / FLUX.1 | ⚠️ Tight | FLUX runs quantised (Q8/Q4) on 8 GB; full precision will not |
| Fine-tuning (QLoRA 7B) | ⚠️ Tight | 7B QLoRA fits with sustained thermals; 13B/SDXL LoRA want 16 GB |
Key:
- ✅ Yes — works well
- ⚠️ Possible but slow — usable with patience
- ❌ No — hardware limitation prevents this
Full Specifications
| Component | Specification |
|---|---|
| CPU | Intel Core i7-11850H (8C/16T) or Xeon W-11855M |
| CPU Generation | Intel 11th Gen (Tiger Lake H, 2021) |
| RAM | 32 GB DDR4-3200 ECC-capable (4× SO-DIMM, max 128 GB) |
| Storage | 1 TB NVMe Gen 4 (2× M.2 2280 slots) |
| GPU | NVIDIA RTX A4000 Laptop (8 GB GDDR6, up to 110W) |
| VRAM | 8 GB GDDR6 (dedicated, ECC-capable) |
| Display | 15.6” 1920×1080 or 4K UHD IPS (config-dependent) |
| Battery | 83 Wh |
| Weight | 2.5 kg |
| TDP | 45W CPU + up to 110W GPU |
| AI Score | 80/100 |
Mobile vs desktop: the A4000 8 GB explained
The desktop RTX A4000 has 16 GB. The RTX A4000 Laptop GPU in the Fury has 8 GB — same name, half the memory, a different chip and a lower power ceiling. This is the single most important thing to understand before buying any workstation laptop: the desktop and mobile parts share a marketing name but not the silicon, and VRAM is what decides whether a model loads. 8 GB is excellent for a portable — it matches the consumer RTX 3080 Laptop and clears SDXL and quantised FLUX — but it is not the 16 GB the desktop card carries. For true 16 GB mobile VRAM you need the ThinkPad P15 Gen 2 and its A5000. We explain the whole trap in what VRAM is and why it matters for AI.
AI Performance in Practice
The RTX A4000 Laptop is a professional Ampere GPU: 3072 CUDA cores, ECC-capable 8 GB, and a power ceiling up to 110W. Raw diffusion throughput sits a little behind the higher-clocked RTX 3080 Laptop — SDXL at roughly 12–18 seconds per 1024×1024 image — but the Fury’s advantage is that it holds that figure across a long batch where a thin laptop would throttle. That makes it the better machine for queued generation and overnight jobs.
For LLMs, 7B runs fully on GPU (an estimated 42–52 tok/s in Q4) and 13B Q4 fits in 8 GB at moderate context — full details in our Ollama laptop requirements guide. FLUX.1 runs quantised via GGUF; see GGUF quantization explained for which quant fits 8 GB. As with every 8 GB card, full-precision FLUX and comfortable fine-tuning want 16 GB.
Thermal behaviour
This is the Fury’s reason to exist. The dual-fan workstation cooler is built to dissipate 45W CPU plus 110W GPU continuously, so the A4000 sustains near its power limit through long training and batch-generation runs without the clock decay that affects thin chassis. It is audible under load but lower-pitched than a gaming laptop, and ISV validation means the thermal solution is tuned for stability rather than peak benchmark scores.
Battery life under AI load
The 83 Wh battery gives 5–7 hours of light work but, like every dGPU workstation, drains in under 90 minutes under sustained GPU load — and the GPU throttles on battery regardless. The 200W slim-tip charger is heavy. Treat the Fury as a mains-powered workstation for AI; use battery for coding and API tools.
What to Check Before Buying (Used)
Confirm the A4000 8 GB — not the T1200/A2000 The Fury 15 G8 shipped with GPUs from the T1200 (4 GB) and A2000 (4 GB) up to the A4000 (8 GB) and A5000 (16 GB). For the AI value here you want the A4000 and 8 GB — or pay up for an A5000 16 GB unit if you can find one. Verify the exact GPU and VRAM in GPU-Z and demand a screenshot; a 4 GB unit is a far cheaper, weaker machine.
RAM slots and capacity The 4× SO-DIMM layout is valuable. Confirm the installed capacity (32 GB common) and that all four slots function — 128 GB is achievable for large-dataset CPU work.
GPU thermals after 3 years Workstation GPUs are robust, but verify. Run FurMark for 15 minutes; a healthy A4000 holds below ~85°C near 110W with no throttling.
Storage and charger Two M.2 Gen 4 slots — check SSD health (CrystalDiskInfo) and consider a second drive for weights. Confirm the correct HP slim-tip 200W charger is included; lower-wattage adapters will not sustain GPU load.
Where to Buy in the UK
Back Market UK — The most reliable graded source, though A4000 8 GB Fury units are less common (limited stock) — expect £750–£1,000 with a 12-month warranty. Confirm the A4000 and 8 GB explicitly.
Laptops Direct — Occasionally lists ex-corporate ZBook Fury workstations from £700. Verify the GPU tier carefully; T1200/A2000 4 GB configs are common.
eBay UK — Best prices (£700–£950) and where most units surface, from CAD/engineering fleets. Always demand a GPU-Z screenshot confirming the A4000 and 8 GB.
What to avoid: Any Fury listing that names “RTX/Quadro” without the model and VRAM. The value is the 8 GB A4000 — a T1200 or A2000 unit is a different, far cheaper laptop.
Verdict
AI Score: 80/100 — Pro AI
The HP ZBook Fury 15 G8 is the sustained-workload 8 GB workstation of our line-up. The RTX A4000’s professional Ampere silicon, ECC-capable 8 GB VRAM and a cooler built for continuous load make it the machine to choose when AI work means long batches and light fine-tuning rather than quick bursts. With 32 GB RAM (expandable to 128 GB) and ISV-grade reliability, it genuinely replaces a desktop for many teams at £700–£1,000 used.
The cost is portability: ~2.5 kg, a heavy 200W brick, and the same 8 GB VRAM ceiling that stops full-precision FLUX and comfortable fine-tuning. If you need those, step up to 16 GB.
Buy if: You run sustained GPU workloads — overnight batches, 7B QLoRA, queued SDXL — and value workstation reliability and thermals over weight.
Don’t buy if: You want portability (the ASUS ROG Zephyrus G14 gives the same 8 GB at 1.7 kg) or you need 16 GB for fine-tuning and full-precision FLUX (the ThinkPad P15 Gen 2). Compare the full range in our best used laptops for local LLMs and best used laptops for Stable Diffusion roundups.