HP Z2 Mini G1a Review

The HP Z2 Mini G1a crams a 16-core AMD Ryzen AI CPU and 128GB of RAM into a mini workstation, making it a local LLM monster. The integrated Radeon 8060S graphics are a weak point for gaming or 3D work, but that's not the point. At around $3,734 from B&H, it's a premium tool for AI developers who need tons of memory on their desk. If you don't need that much RAM, look elsewhere.

Small workstations usually mean compromises, but HP decided to shove a 16-core AMD Ryzen AI Max+ PRO 395 and a frankly ridiculous 128GB of RAM into a box the size of a thick book. The Z2 Mini G1a is aimed straight at developers and data scientists who want to run large language models locally without a server rack humming in the corner. We're talking real AI workloads, not just browser tabs and spreadsheets.

It's not trying to be everything to everyone. HP positioned this as a compact, high-RAM machine for AI-accelerated projects, 3D design, and heavy multitasking. The integrated Radeon 8060S graphics mean you won't be gaming on it, but that's not the point. Our database puts the CPU and RAM at the top of the charts, while the storage and graphics sit more in the middle of the pack. There's a clear priority here: get as much memory bandwidth as possible into a tiny chassis.

The biggest eyebrow-raiser is the price. Across retailers, we've seen numbers ranging from $3,734 all the way up to over $922,000, which has to be some kind of listing error. The real street price from a store like B&H lands around $3,734, and that's the one that actually matters. For a workstation with 128GB of soldered LPDDR5x and a unique AI-tuned processor, that's not outrageous. But you're absolutely paying a premium for the miniaturization.

The star of the show is that 128GB LPDDR5x running at 8533 MHz. In our benchmarks, it lands in the 100th percentile for RAM, which is our fancy way of saying nothing else in this category comes close right now. Couple that with the 3.0 GHz Ryzen AI Max+ PRO 395 and its 16 cores, and you've got a CPU that's comfortably in the top tier, hitting the 91st percentile. For compiling massive codebases, running Docker containers, or feeding a 70B parameter LLM, this thing just chews through it without batting an eye.

The integrated Radeon 8060S is where things get less exciting. It ranks at the 63rd percentile for GPUs in our workstation database, so it's okay, but nothing to write home about. It has dedicated AI acceleration blocks that speed up inference for certain frameworks, but don't expect it to replace a discrete NVIDIA GPU with CUDA. Storage performance is similarly just fine, 64th percentile, thanks to the included HP 1TB PCIe 4.0 SSD. It's enough for booting and project files, but if you're moving around huge datasets, you'll want an external Thunderbolt enclosure or a swap to a higher-endurance drive.

Pricing for the Z2 Mini G1a is a bit of a mess depending on where you look. Some vendors are clearly listing placeholder MSRPs, with one climbing past $922,000. The actual price you'll pay is around $3,734 if you buy from a reputable shop like B&H. That's not cheap, but you're getting a truly unique hardware combination. For context, building a comparable mini-ITX system with 128GB of DDR5 and a similar CPU would run you close to that, and you'd still lack the AI-specific silicon and the polish of a pre-built workstation.

The value argument hinges entirely on your need for gobs of soldered memory in a tiny volume. If you're a developer who wants to run Mixtral or Llama locally while keeping desk clutter to a minimum, the price-per-idea is solid. But if you don't need the bleeding-edge RAM capacity, you can grab a Lenovo Legion Tower 5i for less money and get a far more powerful discrete GPU for rendering or gaming.

Stack the Z2 Mini G1a against traditional tower workstations like the Lenovo Legion Tower 5i or the MSI EdgeXpert, and the trade-offs become clear. Those competitors stuff in proper dedicated GPUs like an RTX 4070, which will wipe the floor with the Radeon 8060S in anything 3D or CUDA-accelerated. They also have room for more storage and internal expansion. The HP sacrifices all that for the sake of being the size of a hardcover novel and having the absolute best RAM setup right now.

Then there's the ASUS Republic of Gamers GM700TZ, which is built for high-refresh-rate gaming and comes with a muscular NVIDIA card. If you're even a little bit interested in gaming, that's a smarter buy. The Dell XPS EBT2250 is a more pedestrian competitor that doesn't quite match the HP's AI chops but offers a more balanced mix of CPU and GPU power in a larger enclosure. So the HP wins when your workflow screams for memory bandwidth and AI inference, and loses practically everywhere else.

Skip the Z2 Mini G1a if you're a 3D artist, video editor relying on GPU acceleration, or anyone whose workflow requires a powerful discrete GPU. The integrated Radeon 8060S just can't compete with an RTX card for rendering or real-time 3D viewports, and you'll notice the lag immediately. Also, if you need a dead-silent machine, the small chassis means the fans can get loud when the CPU is pinned. Anyone invested in CUDA-specific libraries will be better served by a tower like the MSI EdgeXpert or a Dell XPS with an NVIDIA GPU. And finally, if 128GB of RAM sounds like overkill, you're paying a steep premium for silicon you'll never use, so grab a cheaper mini PC with a more balanced spec sheet.

If you're a machine learning engineer who wants to experiment with large language models without touching the cloud, the HP Z2 Mini G1a is a no-brainer. That 128GB of RAM means you can load full-precision 70B models into memory and still have room for your IDE and a hundred browser tabs. The CPU is fast enough that local fine-tuning with smaller models is totally feasible, and the tiny footprint means you can stash it behind a monitor. As a developer tool, it's exceptional.

For everyone else, the GPU bottleneck is real. 3D artists and game developers will find the Radeon 8060S limiting, and anyone with a gaming hobby will be disappointed. If your work involves heavy GPU rendering or you need CUDA for your tools, just get a tower with a dedicated NVIDIA card. The Z2 Mini G1a nails one specific job, and that's being an AI powerhouse in the smallest possible box. It's laser-focused, and if you're in that niche, you'll love it.