Minimum System Specifications
- RAM: 2GB
- CPU: 2-core (any architecture, supports Raspberry Pi)
- Storage: 5GB
Component Overhead & Scaling
1. Large Language Model (LLM)
- Cloud/External APIs (e.g., OpenAI): Near 0% local resource overhead. Requires API key.
- Local LLM: If running on the same machine, requires dedicated GPU. If the host lacks a GPU, run the LLM on a separate GPU-enabled machine and connect AnythingLLM via API.
2. Embedder (Vectorization)
- Default Embedder: Runs locally on CPU. Resource consumption scales with document size and ingestion volume.
- External Providers: Near 0% local resource overhead. Recommended for large-scale data ingestion.
3. Vector Database
- Default (LanceDB): Local embedded database. Runs within default system specifications.
- External Databases: Near 0% local resource overhead. Scales to hundreds of millions of vectors.