AES-256-GCM encryption with Argon2id key derivation. CMMC 2.0 Level 2 certified for defense contractors.
SafeTensors, GGUF, PyTorch, ONNX, TensorRT, MLX, Core ML, TFLite, OpenVINO, HDF5, and more. Auto-detected.
Complete version history with lineage tracking, branching for A/B testing, and instant rollback to any checkpoint.
8 built-in tools: archive, deduplication, analysis, export, caching, quantization tracking, pruning, and compression.
Document store with vector embeddings, semantic search, MCP protocol support, and 4 built-in RAG tools.
Windows, Linux, macOS. 100% XDG Base Directory compliant. 15+ CLI commands and full Rust API.
# Store a model with encryption
aim store llama-3-8b.gguf --encrypt --tags "llm,production"
# Retrieve a model
aim get llama-3-8b --version latest
# List all stored models
aim list --format table
# Archive for backup
aim archive llama-3-8b --format tar.gz
# Deduplicate storage
aim dedup --dry-run
# Version history with lineage
aim history llama-3-8b --show-lineage