Local AI Hardware
Mini PCs, unified-memory systems and DIY builds for running language models locally.
23 items · manufacturer-sourced
See the ranked shortlist →Hardware

Mac Studio (M3 Ultra)
Apple
9.8ScoreEMC8859—HDMIHDMI 2.1
Mac Studio M2 Ultra
Apple
9.8ScoreGPU76-core GPUEnthusiastMemoryUp to 192GB Unified
Nvidia GeForce RTX 5090 (32GB)
Nvidia
9.7ScoreMemory32GB GDDR7—Cuda Cores21760
Mac Mini M4 Pro (64GB)
Apple
9.5ScoreUnified memory64 GBSweet spotThroughput35-50 tok/s
NVIDIA GeForce RTX 4090
NVIDIA
9.5ScoreTdp450WEnthusiastHDCP2.3
Nvidia GeForce RTX 5070 Ti (16GB)
Nvidia
9.3ScoreMemory16GB GDDR7—Cuda Cores8960
ClawBox
ClawBox
9.2ScoreGPUNVIDIA Ampere—RAM8GB LPDDR5 unified memory
Minisforum MS-S1 Max
Minisforum
9.2ScoreN5AMD Ryzen™ 7 255 / 16G DDR5 + 128G OS Storage—APUAMD Ryzen AI Max+ 395
Threadripper Pro + Dual RTX 4090 Build
9.2ScoreTotal VRAM48ProfessionalRAM256 GB
Nvidia GeForce RTX 5080 (16GB)
Nvidia
9.1ScoreMemory16GB GDDR7—Cuda Cores10752- MinisforumImage unavailable
Minisforum MS-01
Minisforum
8.5ScoreForm factormini-pcSweet spotCores14 cores and 20 threads 
Nvidia GeForce RTX 3090 (Used)
Nvidia
8.5ScoreBoost Clock1.86 GHz | 1.70 GHz—Memory Size24 GB | 24 GB
GMKtec EVO-X2 (128GB)
GMKtec
8.0ScoreUnified memory128 GBEnthusiastThroughput15-25 tok/s
Ryzen 5 + RTX 4060 Ti Build
7.5ScoreVRAM16 GBEntryRAM64 GB- NVIDIAImage unavailable
GeForce RTX 5060
NVIDIA
FP4306.8 TFLOPS—FP6153.4 TFLOPS - NVIDIAImage unavailable
GeForce RTX 5060 Ti
NVIDIA
FP4379.3 TFLOPS—FP6189.6 TFLOPS - NVIDIAImage unavailable
GeForce RTX 5070
NVIDIA
FP4493.9 TFLOPS—FP6246.9 TFLOPS - NVIDIAImage unavailable
NVIDIA DGX Spark 64GB
NVIDIA
ComputeNVIDIA Grace Blackwell—NetworkingNVIDIA ConnectX-7 
NVIDIA RTX Spark Windows PCs
NVIDIA
DLSSedit 12K 4:2:2 video with the NVIDIA Blackwell decoder—Size14- to 16-inch sizes- AMDImage unavailable
Radeon RX 9060 XT
AMD
FP8205.1 TFLOPS—TDP160 W - AMDImage unavailable
Radeon RX 9070
AMD
Rops128—Density151.0M / mm² - AMDImage unavailable
Radeon RX 9070 GRE
AMD
TDP220 W—TPP2194 - AMDImage unavailable
Radeon RX 9070 XT
AMD
FP8389.3 TFLOPS—FSR4.1.2, XeSS 3
Further reading
- guideHow to build a local AI server with used GPUs: the RTX 3090 & Tesla P40 guide
- guideRunning and fine-tuning open models on Windows (WSL2, DirectML, CUDA)
- guideOptimising local LLM inference: where the bottleneck actually is
- guideBuilding a Tesla P40 rig for local AI: what you're really signing up for
