✦

AI Studio — Local Model Hub

Interactive mock of the Android app (Kotlin + Jetpack Compose) · models from registry.json, updated 2026-09-12

Backend: …

Device free RAM GB

Backends

Ollama base URL (Backend.Ollama default)

Router (LocalModelRouter): prefer a reachable Ollama server; otherwise stay on device.

Served by Ollama — GET /api/tags

On-device fallback

Chat

Models use–

Installed–

Performance modeBalanced

Max context / parallel–

Storage / Performance

Memory mode (MemoryAndModes.PerformanceMode)

Preview mock — no real inference or downloads. Compatibility mirrors CompatibilityChecker (weights ×1.2; “fits” ≤ 70% of free RAM), routing mirrors LocalModelRouter, modes mirror PerformanceMode (Battery 2048/1 · Balanced 4096/1 · Performance 8192/2). Source: preview/index.html in this project.