Free · No sign-up · Works in your browser
Free AI hardware tools and calculators
Everything you need to size, compare and understand AI hardware: calculate the VRAM a model needs, estimate training time and cost, compare GPUs side by side and see exactly how AI number formats work. Every tool uses published specifications and shows its formulas.
Calculator
LLM VRAM calculator
How much GPU memory does Llama, Qwen, DeepSeek or gpt-oss need? Weights, KV cache, GPU fit and tokens per second.
Open →Calculator
AI training time and cost calculator
Compute, GPU-hours, days, cloud cost and energy to train a model, using 6 × parameters × tokens.
Open →Data
AI GPU specs database
28 accelerators from NVIDIA, AMD, Intel, AWS and Google compared per chip, with sorting, head-to-head and CSV.
Open →Calculator
AI model GPU calculator
Memory, Tensor Core work per token and speed for popular models on H100, H200, B200 and B300.
Open →Converter
Number format converter
See how FP32, TF32, BF16, FP16, FP8 and FP4 store any number, bit by bit.
Open →Reference
Tensor Core facts and statistics
25 verified, sourced facts with individual links for citing.
Open →Games
TensorCore.si Arcade
Five two-minute games on matrix math, number formats, the roofline model, GPU history and GPU cores.
Open →Challenge
Tensor Core math challenge
17 randomized problems in three levels. Score 80% to download a verifiable certificate.
Open →Reference
AI hardware glossary
Plain-language definitions of every Tensor Core, GPU and AI hardware term, A to Z.
Open →Which tool do I need?
| Your question | Use |
|---|---|
| Can I run this model on my GPU or Mac? | LLM VRAM calculator |
| How long and how much to train a model? | Training time and cost calculator |
| Which data-center GPU has the most memory or FLOPS? | AI GPU specs database |
| How many H100s or B200s does a model need? | AI model GPU calculator |
| What does FP8 or FP4 do to my numbers? | Number format converter |
| I need a sourced statistic for an article | Tensor Core facts |
New to the topic? Start with what is a Tensor Core and how to read a GPU spec sheet.