All notable changes to Memvid will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Initial public release of Memvid core library
- Single-file
.mv2format for portable AI memory - Full-text search with BM25 ranking (Tantivy)
- Vector similarity search with HNSW
- PDF, DOCX, XLSX document ingestion
- CLIP visual embeddings for image search
- Whisper audio transcription
- Timeline queries for chronological browsing
- Crash-safe WAL-based writes
- Blake3 checksums for data integrity
- Ed25519 signatures for authenticity
- Optional AES-256-GCM encryption
- Embedded WAL prevents data corruption
- Atomic commits ensure consistency
- File locking prevents concurrent write conflicts
2.0.0 - 2026-01-05
- Complete rewrite in Rust for performance and safety
- New
.mv2file format (single-file, no sidecars) - Append-only frame-based architecture
- Built-in full-text and vector search
- Cross-platform support (macOS, Linux, Windows)
- Migrated from Python to Rust
- New API design focused on simplicity
- Improved memory efficiency
- Legacy Python implementation
- QR code video encoding (replaced with efficient binary format)