About
RunAIatHome
Helping everyone run AI on their own hardware. Real benchmarks, no fluff.
RunAIatHome Team
Building local AI tools since 2023
We wanted to run AI at home in 2023 but couldn't find a single good resource that answered "what can my GPU actually run?" — not vague estimates, actual numbers. So we built it.
RunAIatHome started as a spreadsheet with benchmark data from real hardware. Now it's a proper tool. Every number you see here comes from real inference runs, not manufacturer spec sheets.
Hardware Lab
Our Mission
We believe that everyone should have access to powerful AI tools without depending on cloud services, monthly subscriptions, or giving up their privacy. RunAIatHome exists to make local AI accessible to everyone, regardless of their technical background.
Running AI at home used to require deep technical knowledge and expensive hardware. That is changing fast. Modern open-source models and tools have made it possible to run capable AI on consumer hardware. Our job is to help you navigate that landscape.
Todos los datos de tokens/segundo y compatibilidad de modelos publicados en este sitio provienen de ejecuciones reales en nuestro laboratorio. No usamos hojas de especificaciones de fabricantes ni simulaciones.
— RunAIatHome Hardware Lab — AMD Ryzen 9 7950X, 128GB DDR5, NVIDIA driver 550Mantener los datos actualizados es parte del compromiso editorial de RunAIatHome. Cuando un modelo importante sale, lo probamos y publicamos resultados antes de que la mayoría de recursos lo documente.
— Política editorial de RunAIatHomeWhat We Do
Hardware Assessment
Our AI Hardware Wizard analyzes your current setup and tells you exactly what AI models you can run, what upgrades would help, and what to buy.
Free Tools
Our tool suite includes a VRAM calculator, GPU comparator, model browser, build configurator, and cost calculator. All free, no sign-up required.
Guides & Tutorials
Clear, practical guides that help you go from zero to running AI locally. From installing Ollama to choosing the right GPU.
Hardware Recommendations
Curated hardware catalog with components tested and selected specifically for AI workloads. Real benchmarks, honest recommendations.
How We Test
Real Hardware
All benchmarks run on physical hardware. No simulations, no interpolations. RTX 4090 for high-end GPUs, RTX 3060 12GB to validate typical consumer hardware.
Documented Conditions
Every result includes the full environment: NVIDIA driver 550, CUDA 12.4, Ollama 0.5.x, Ubuntu 24.04 LTS. GPU temp < 80°C. No overclocking.
Primary Metric: tokens/second
Measured with 512-token prompt, 256-token response. Average of 3 runs, discarding the first (warm-up). Tools: ollama bench + custom script.
Monthly Updates
When a new model ships, we benchmark it within 72h. Stale data = broken tool — we take this seriously.
Found an error in our data? Email hello@runaiathome.com — we fix it the same day.
Our Values
Privacy First
We advocate for running AI locally precisely because of privacy. We practice what we preach: minimal data collection, no tracking beyond basic analytics, no selling user data.
Honest Recommendations
We only recommend hardware we would buy ourselves. Our recommendations are based on real benchmarks and practical experience, not marketing materials.
Free & Accessible
All our tools and guides are free. We sustain the project through affiliate commissions on hardware we recommend. You never pay more because of us.
Editorial Policy
Affiliate links
Some links on this site are affiliate links (Amazon Associates). We only link to hardware we've tested or would personally buy. Affiliate commissions never influence our recommendations — if a lower-tier GPU does the job, we will not push you toward a premium card.
Data accuracy
Hardware data is sourced from our own tests, manufacturer specs (verified), and reputable benchmarks. All data includes source and test date.
No paid placements
We do not accept payment for reviews, rankings, or mentions. No sponsored content. Ever.
Affiliate Disclosure
RunAIatHome participates in the Amazon Associates Program and other affiliate programs. When you click on a product link and make a purchase, we may earn a small commission at no additional cost to you.
This affiliate revenue is what allows us to keep all our tools and guides free. We appreciate your support.
Important: Affiliate relationships never influence our recommendations. We recommend products based on benchmarks, specifications, and real-world testing. If a product is not good for AI, we will not recommend it regardless of affiliate potential.