Local AI / No guesswork

Know what your machine can run.

Practical RAM, VRAM, and storage estimates for popular open models—calculated from live Hugging Face metadata.

Check your hardware →Compare every model
● Build-time Hub data● No runtime tracking● Open methodology

Hardware fit preview

32 GB RAM · 8 GB VRAM

Live estimate
Llama 3.2 3BQ8_0
Comfortable
Mistral 7BQ4_K_M
Comfortable
Qwen 14BQ4_K_M
Tight
Llama 70BQ4_K_M
Not advised
VRAM allocated6.1 / 8 GB

Popular models

The short list.

Explore all configurations →
ModelParametersQ4 VRAMCPU RAMDisk
Qwen 2.5-3B ↗3.1B3.5 GB3.9 GB1.8 GB
Phi 3.5 Mini ↗3.8B4 GB4.5 GB2.2 GB
Mistral 7B ↗7.2B6.2 GB7.2 GB4.1 GB
Hermes-3-Llama-3.1-8B ↗8B6.7 GB7.8 GB4.6 GB
Qwen 2.5-14B ↗14.8B11.2 GB13.1 GB8.4 GB
Qwen 2.5-72B ↗72.7B49.2 GB58.7 GB41.4 GB

Three useful answers

From specs to a confident choice.

01

Compare models →

See equivalent FP16, 8-bit, and 4-bit footprints side by side.

02

Match your hardware →

Enter what you own and get realistic fit bands in seconds.

03

Choose a laptop →

Understand where unified memory, dGPUs, and RAM tiers matter.

Ready when you are

Stop buying hardware by model-name hype.

Start with the memory you have—or the model you want—and work backward from the numbers.

Find your fit →