Local AI Fit

Local model compatibility

Know before you download.

Check whether a local AI model fits your machine, what to install, and where the estimate comes from. Built for people who would rather test than guess.

Check my machine Browse models

Compatibility check

Can your machine run it?

Pick a machine, a model, a quantization, and a context length. The result uses published weight sizes plus a conservative KV-cache estimate and 1.5 GB runtime overhead. It is not a benchmark.

Coverage

Recent models people actually ask about.

All models
MoE

gpt-oss-20b

OpenAI · Apache 2.0

21B total / 3.6B active · 131,072 context

From 12.11 GB · consumer

MoE

Qwen3.6 35B-A3B

Alibaba Cloud · Apache 2.0

35B total / 3B active · 262,144 context

From 22.61 GB · workstation

Dense

Qwen3.8 27B

Alibaba Cloud · Apache 2.0

27B dense · 262,144 context

From 17.39 GB · workstation

Dense

Gemma 4 12B Unified

Google DeepMind · Apache 2.0

11.95B dense · 262,144 context

From 7.15 GB · consumer

Curious whether the 552B DeepSeek V4.1 Flash can run locally? Read the server-tier reality check.

Hardware

Compare the machine you have.

All hardware

Method

Why the numbers can be trusted.

Every model page lists its source and last-verified date. Minimum memory is a conservative starting point, not a guarantee. If a number is unknown, the page says so instead of inventing precision.

The checker does not send your selections anywhere. It runs in your browser.