Local Ollama
A good fit for Macs, desktops, capable NAS devices, or VPS boxes. Multilingual embedding models with 1024 dimensions are a practical starting point for Chinese, Japanese, and English RSS.
Local embedding keeps vector generation inside your own machine.
Compatible providers
SiliconFlow, Gemini AI Studio, and other OpenAI-compatible endpoints can generate embeddings. The provider creates vectors; ranking, profile, explanations, and index state remain inside your instance.
Failures are captured as diagnostics and graceful fallback, not as a broken reader.
Index maintenance
Active indexes can backfill missing or stale embeddings. sqlite-vec can rebuild from the local authority table without calling the provider again.
Diagnostics expose coverage, candidate size, pending/failed jobs, clusters, and warnings without leaking API keys or raw vectors.