Lenovo ThinkStation P3 Tiny Gen 2

A seriously dense little workstation with a 20-core Intel CPU and 64GB of RAM that ranks in the top tiers of our benchmark database. The 8GB NVIDIA RTX GPU is mid-pack, so it’s built for professional apps, not gaming. At $1,459 without an OS, it’s a premium compact tool best suited for developers and engineers who need Windows or Linux in a vanishingly small chassis. If you need GPU horsepower or are on a budget, look elsewhere.

You know that feeling when you need real workstation muscle but you're completely out of desk space? Lenovo’s ThinkStation P3 Tiny Gen 2 is basically the answer to that dilemma. It crams a 20-core Intel Core Ultra processor, 64GB of DDR5 RAM, and a discrete NVIDIA RTX GPU with 8GB of VRAM into a 1.4-kilo box that’s about the size of a thick paperback. No OS is preinstalled, so you're free to slap on Windows 11 or your favorite Linux distro and get to work.

We see a lot of tiny PCs, but most of them trade away serious compute for the sake of size. This one doesn’t. The 64GB RAM config puts it in the top 3% of all machines we’ve tested in its class, and that 20-core CPU sits comfortably in the 89th percentile. It’s aimed straight at developers, engineers, and IT pros who need a quiet little box that can compile massive codebases, run VMs, or handle CAD workloads without taking over the whole desk.

What makes it interesting is the sheer density. You’re getting workstation-level components in a 1-liter chassis that can be mounted behind a monitor or stuffed into a server rack shelf. And the optional punch-out ports mean you can tailor the back-panel connectivity to your exact display setup, from legacy VGA to Thunderbolt 4. If your job demands a capable x86 machine that disappears into your workspace, this thing deserves a long look.

Our benchmark database shows the Core Ultra 20-core chip here is no slouch, landing in the 89th percentile for CPU-bound tasks. In multi-threaded workloads like code compilation or data analysis, it’s faster than a lot of full-sized desktops we’ve tested. Paired with 64GB of DDR5 (97th percentile), it chews through RAM-hungry projects without breaking a sweat. We ran our usual V-Ray and Cinebench loops, and it consistently outpaced last-gen Core i9 laptops and even some older tower workstations.

The GPU story is more nuanced. That 8GB RTX card sits at the 58th percentile, which means it’s right in the middle of the pack. For professional apps like SolidWorks, Revit, or light Blender work, it’s perfectly fine. But if you push it with GPU rendering or try to game on it, you’ll feel the limits fast. Our gaming score for this config came in at 61.6 out of 100, so it really isn’t built for play. The 1TB NVMe SSD (73rd percentile) is quick enough for daily use, though sequential transfers aren’t chart-topping. The 170W external PSU keeps everything fed, but the small fans do spin up noticeably under sustained load, so you’ll want it at arm’s length if you’re sensitive to fan noise.

At $1,459 with no OS, the P3 Tiny Gen 2 is a tough call purely on price-to-performance. For the same money you could build a small ITX rig with a much stronger GPU and similar CPU grunt, though it wouldn’t be nearly as compact. On the other hand, Apple’s Mac mini M4 starts at just $599 and often beats this in single-core speed and silence, but it tops out at 32GB of RAM and lacks a discrete GPU (and runs macOS, which might be a dealbreaker depending on your workflow).

If you’re a developer or IT pro who genuinely needs 64GB of RAM, Windows or Linux, and a CUDA-capable GPU in the smallest possible footprint, the Lenovo makes a lot of sense. You’re paying a premium for that density and the ThinkStation reliability. If you can live with a slightly larger chassis or don’t need the GPU, there are more affordable paths, but as a purpose-built tiny workstation, it holds its own.

The most direct rival is the Apple Mac mini M4. The M4 chip’s single-core performance is outstanding and the machine is dead quiet. But you’re locked into macOS, the RAM ceiling is lower, and there’s no dedicated NVIDIA GPU for CUDA workloads. The P3 Tiny gives you x86 flexibility, a ton of RAM, and that discrete RTX card, which matters for certain engineering and rendering apps. On the flip side, the Mac mini is cheaper by a wide margin and more power-efficient.

Then there are the gaming-focused towers like the HP OMEN 45L or ASUS ROG GM700TZ. Those machines humiliate this little Lenovo in GPU tests, with RTX 3080-class cards pushing gaming scores above 90 in our database. But they’re massive, heavy, and complete overkill if you just need a workstation. The Dell XPS desktop (EBT2250) sits somewhere in the middle, but it’s still a tower. If you want a true sub-2kg workstation that can live behind a monitor, you’re really choosing between this ThinkStation and something like an Intel NUC with a weaker GPU. The Lenovo wins that battle on core count and RAM capacity.

Gamers should absolutely skip this machine. With a gaming score of 61.6, it’s far behind dedicated gaming PCs in its price range, and you can get a compact tower with an RTX 4060 or better for similar money that will run circles around it in games. If you’re into AI model training or heavy GPU compute, the 8GB VRAM cap is a hard no, look for at least 12GB or 16GB cards in a larger workstation.

Anyone on a strict budget who doesn’t need Windows or Linux should also think twice. The Mac mini M4 base model costs less than half as much and delivers superior single-core speed and silent operation, though with less RAM and no CUDA support. And if you’re a creative pro doing 4K video editing all day, the lack of an OS and middling GPU mean you’d be better served by an M2 Pro Mac mini or a Dell Precision tower with a faster Quadro card.

For software developers, Linux admins, and engineers who need a dense block of x86 compute that vanishes into a clean desk setup, the P3 Tiny Gen 2 is one of the best options right now. The 20-core CPU and 64GB of RAM let it handle virtual machines, heavy compilation, and CI/CD tasks without choking, and the discrete GPU gives it a leg up over pure iGPU mini PCs. If your daily work involves CAD or moderate CUDA-accelerated tasks, it’ll slot in nicely.

If your workload leans heavily on GPU rendering, AI model training, or you just want to game after hours, this isn’t the machine for you. The 8GB RTX card is a bottleneck, and the form factor simply can’t dissipate the heat of a faster GPU. In that case, you’re far better off with a larger desktop tower, even if it means sacrificing some desk real estate. And if you can live inside macOS and don’t need more than 32GB of RAM, the Mac mini M4 is a serious value alternative that’ll save you a chunk of cash.