Public calculator
GPU to LLM sizing is visible before any sales contact: VRAM, quantization, single GPU and multi-GPU rigs.
LocalIA is built by Damien Ogryzek. The goal is simple: help freelancers, SMEs and technical teams run open-weight LLMs locally, with readable sizing before purchase.
GPU to LLM sizing is visible before any sales contact: VRAM, quantization, single GPU and multi-GPU rigs.
Model and GPU pages are built from versioned data and a daily Hugging Face sync.
The site is published by Damien OGRYZEK, Entrepreneur individuel (micro-entreprise), SIREN 918508862.
No opaque checkout: email scoping, firm quote, deposit, announced build time, then insured delivery.
The calculator can bring traffic. To sell a machine worth several thousand euros, the site also has to be clear about its current proof level.
You send the target model, users, data constraints, GDPR concerns and approximate budget.
I compare VRAM, CPU, RAM, storage, power, noise and upgrade margin, then propose a config.
You receive a VAT-included price, estimated timeline, performance assumptions and warranty terms.
After validation, the rig is assembled, installed, tested, documented and delivered with onboarding.
The next LocalIA step is not more advertising: it is photos, a short video and benchmarks on a physical machine. Until then, performance remains sizing estimates.
Test the calculator ->