A self-hosted AI platform,
running my own apps.
Instead of paying per-token for a proprietary API, every app I build talks to Yuu v1.1 — an open-weight model (Qwen3-8B) I run behind my own OpenAI-compatible gateway, with its own auth, rate limiting, and usage tracking.
How it works
One gateway
A single API (OpenAI-compatible) fronts the model, with its own API keys, per-app rate limits, and cost tracking.
Every app plugs in
Each app has its own dedicated key and its usage is tracked separately — see the dashboard for real numbers.
No vendor lock-in
The underlying model is swappable through config — today it's Qwen3-8B, tomorrow it could be anything else.
Featured projects
View all →
Nodability
AI-chat-driven personal task organizer for a couple, with task board and calendar views.
Engo
Video-first language-learning platform teaching Japanese/English through real YouTube lessons.
FLUENT
3D Tokyo life-sim for learning Japanese, with AI-driven NPC conversations.
LeanGraph
Benchmark of LLM agents writing Lean 4 proofs, where a proof only counts if Lean verifies it.