v1.0.0
Released: 2026-05-04 Changes:
Initial release. Three services with OpenAI-compatible API: VLM (Gemma 4 E4B), Embedding (BGE-base-zh), STT (Whisper large-v3-turbo). All bind to 127.0.0.1. Installer ~250MB; first launch auto-downloads ~6.3GB AI models from HuggingFace (10-30 min).