The independent guide to running AI locally.
Detect your GPU in the browser, see which open-weights models actually fit your VRAM, and calculate what self-hosting saves versus commercial APIs. No sign-up, nothing installed, everything stays in your browser.
- WebGPU + WebGL detection with manual hardware selection fallback
- Recommendations matched to your real memory budget
- Savings calculator: own API price when it exists, otherwise closest commercial equivalent
- Directory pages for every model, built for search engines
1. Detect your GPU
WebGPU GPUAdapterInfo + WebGL WEBGL_debug_renderer_info. The browser reports the chip, never the memory size — we map the chip to a hardware database.
WebGPU
Checking…WebGL
Checking…Selected hardware
No hardware selected yet — pick one below.Recommended for your hardware
Detect or select hardware to see recommendations.
Self-hosting savings
prototype pricing · 2026-08Waiting for hardware detection…
Featured models
Full directory →StepFun
Step 3.7 Flash
148B params · Intelligence 78.8
DeepSeek
DeepSeek V4 Flash 0731
162B params · Intelligence 77.5
DeepSeek
DeepSeek V4 Flash
162B params · Intelligence 77.3
Qwen
Qwen3.6 27B
27B params · Intelligence 76.5
Qwen
Qwen3.6 35B A3B
35B params · Intelligence 74.9
HCompany
Ornith 1.0 35B
35B params · Intelligence 74.1
What this site is becoming
A decision engine, not just a benchmark table: model detail pages, VRAM guides, comparisons, and hardware-buying advice that ends in a purchase decision. The content roadmap is in content/SEO_CONTENT_PLAN.md; the product and monetization strategy is in docs/PLAN.md.