# AI Hardware Fit

A free, browser-based calculator, in Korean and English, that estimates which local AI models a given GPU can run, or which GPUs a chosen model needs, along with quantization and speed estimates, with no install, sign-up or server-side processing.

AI Hardware Fit lets a user either pick a GPU and see which local language, vision-language, embedding or reranking models it can run along with a recommended quantization and estimated speed, or start from a model and compare candidate GPUs against a VRAM, budget or power constraint. A separate infrastructure-sizing mode turns a described service, such as an internal document RAG system or a customer chatbot with a given number of users and concurrent requests, into a suggested server configuration, a total cost of ownership estimate and a proof-of-concept checklist.

All calculations run inside the browser rather than on a server, and the project states plainly that it does not provide guaranteed automatic quotes or performance guarantees, intending it to support pre-sizing, comparison and hypothesis-building ahead of a real proof of concept. Additional calculators cover estimated API costs and the one-time cost of structuring a set of documents into an ontology or knowledge graph using an LLM.

It is written in JavaScript and released under the MIT license.

Website: https://jaeseok614.github.io/llm-gpu-checker-ko/