Fit
Model, context and reserve memory are checked against your available memory.
Find the right local LLM
Match models to your hardware and workload.
Matches
Example: 16 GB RAM.
Choose hardware and a main use.
How it works
How the finder works
Rank
Models are ranked for your selected workload and speed/quality preference.
Compare
Known GPUs also get a rough bandwidth-based speed range.
Common questions
Local LLM basics
What is a local LLM?
A language model that runs on your computer. Tools such as Ollama, LM Studio and llama.cpp can run compatible models locally.
How much VRAM do I need?
It depends on model size, quantization and context. Smaller models can use a few GB; larger ones may need 16, 24, 48 GB or more.
Does more RAM or VRAM make a model smarter?
No. More memory lets you run larger models or longer context; it does not improve the same model by itself.
Are speed figures benchmarks?
No. They are rough planning estimates. Runtime, drivers, CPU, settings and prompt length can change real performance.
Before relying on a result
Results are experimental estimates. Check the model publisher, licence and runtime before relying on them or buying hardware.