About xpu.wiki
The comprehensive, vendor-neutral encyclopedia for AI compute hardware
Our Mission
xpu.wiki exists to provide accurate, up-to-date, and unbiased information about AI accelerators. Whether you're a researcher choosing hardware for training, an engineer optimizing inference costs, or a datacenter operator planning infrastructure, we help you make informed decisions.
We believe that transparent, data-driven comparisons benefit the entire AI ecosystem by helping teams select the right hardware for their workloads and budgets.
Features
Comprehensive Hardware Database
Detailed specifications for 40+ AI accelerators from NVIDIA, AMD, Intel, Google, AWS, and more. Includes GPUs, TPUs, NPUs, IPUs, and specialized inference accelerators.
Side-by-Side Comparison
Compare up to 5 XPUs simultaneously across compute performance, memory specs, power efficiency, and pricing. See exactly how many units of GPU A you'd need to match GPU B's performance.
Real-World Pricing Data
Cloud rental prices (OPEX) from major providers like AWS, Google Cloud, Azure, Lambda Labs, RunPod, and Salad. Hardware purchase prices (CAPEX) for on-premise deployments. Updated regularly.
Performance Benchmarks
LLM training and inference metrics, image generation performance, and specialized workload benchmarks. All data sourced from vendors and verified third-party tests.
Vendor & Provider Directory
Learn about hardware manufacturers (NVIDIA, AMD, Intel, Google, etc.) and cloud providers offering GPU compute. Includes company history, product focus, and market positioning.
Educational Encyclopedia
Understand technical terms like TFLOPs, HBM, precision types (FP32, BF16, FP8), form factors (PCIe, SXM, OAM), and more. Written for engineers and decision-makers.
Brought to you by hosted·ai
xpu.wiki is created and maintained by hosted·ai, a platform helping businesses find and deploy AI compute infrastructure.
At hosted·ai, we help companies navigate the complex landscape of AI hardware and cloud providers. We built xpu.wiki as a free resource for the community to make better informed hardware decisions.
Data Sources & Accuracy
We gather data from official vendor specifications, cloud provider pricing pages, and verified benchmark publications. All sources are linked for transparency.
Pricing: Cloud pricing updated weekly. Hardware CAPEX prices represent typical street prices, not MSRP, and vary by volume and region.
Benchmarks: Performance metrics from official vendor benchmarks and reputable third-party testing. Real-world performance varies by workload, software stack, and configuration.
Help Improve xpu.wiki
Found incorrect data? Have benchmark results to share? Want to see a specific GPU added?
We welcome contributions from the community. Reach out to help make xpu.wiki more comprehensive and accurate.
Get in Touch