Devtoolwatch BoardMemvid review
Memvid is a single-file memory layer for AI agents that packages data, embeddings, search indexes, and metadata into one portable .mv2 file.
TL;DR
Memvid is a portable memory system that stores data, embeddings, indexes, and metadata in one .mv2 file. Agents retrieve directly from that file, with no database or server. It is for developers building AI agents that need persistent, offline-capable memory. Its append-only Smart Frames design supports queries over past memory states.
- 17.3kGitHub stars
- +731new stars, week of Oct 25
- #2in RAG, data and memory this week
Memvid at a glance
- Category
- RAG, data and memory
- Maintained by
- Memvid (organization, @memvid on GitHub)
- License
- Apache 2.0
- Language
- Rust, Shell, Makefile
- Repository
- github.com/memvid/memvid
- Website
- www.memvid.com
- Latest release
- v2.0.140 on GitHub, May 27, 2026
- First published
- May 27, 2025
What Memvid does
- Memvid stores content, embeddings, a full-text index, a vector index, and a time index in a single .mv2 file with no sidecar files.
- Memvid offers full-text search with BM25 ranking through Tantivy and vector similarity search with HNSW, enabled as feature flags.
- Memvid writes memory as append-only, immutable Smart Frames with timestamps and checksums, so you can query past memory states.
- Memvid includes an embedded write-ahead log for crash recovery.
- Memvid can extract PDF text, create CLIP image embeddings, and transcribe audio with Whisper through optional features.
- Memvid supports local ONNX text embeddings or OpenAI API embeddings, and can bind an index to one embedding model.
- Memvid supports password-based encryption capsules (.mv2e) and frame-level access control lists.
How it works
Memvid runs as a library inside your application. A CLI and SDKs exist for Node.js, Python, and Rust. You create a .mv2 file, add documents with metadata, commit, and search it directly. The homepage says agents and models can connect through MCP, SDK, or direct API. The file can be stored locally, on-prem, or in a cloud.
How to install and start using Memvid
CLI with npm
npm install -g memvid-cliPython SDK with pip
pip install memvid-sdkRust crate with Cargo
cargo add memvid-coreRequirements
- Rust 1.85.0 or later for the Rust crate
- Local text embeddings need ONNX model and tokenizer files downloaded manually to ~/.cache/memvid/text-models
- OpenAI embeddings need the api_embed feature and an OPENAI_API_KEY environment variable
First steps
- Memvid's Rust crate can be added with features enabled, for example lex, vec, and temporal_track, in Cargo.toml.
- Create a memory file with Memvid::create("knowledge.mv2"), add documents with put_bytes_with_options, then call commit.
- Search the file with mem.search and a SearchRequest that sets the query, top_k, and snippet length.
- Run the bundled example to try create, put, search, and timeline operations: cargo run --example basic_usage
Who Memvid is for
- Developers building long-running AI agents that need persistent memory without running a vector database.
- Offline-first or air-gapped AI systems that must keep data local.
- Teams that need auditable AI workflows with replay and inspection of past memory states.
- Knowledge bases and codebase understanding tools that mix full-text and vector search.
Limitations to know
- Memvid's local text embeddings do not download automatically. You fetch ONNX model files yourself before using the vec feature.
- Memvid binds an index to one embedding model once set, and switching models later fails with a ModelMismatch error.
- Memvid v1, the QR-code based version, is deprecated, so older tutorials referencing QR codes are outdated.
- Memvid recent releases fixed WAL corruption and time index crashes, so staying on current versions matters.
Is Memvid growing? Usage data
In the week of Oct 25–31, 2026, Memvid ranked #29 on the Devtoolwatch Board and #2 in RAG, data and memory, with a momentum score of 84 out of 100.
Show the weekly numbers
| Week of | New stars |
|---|---|
| Aug 9, 2026 | 29 |
| Aug 16, 2026 | 218 |
| Aug 23, 2026 | 25 |
| Aug 30, 2026 | 39 |
| Sep 6, 2026 | 55 |
| Sep 13, 2026 | 18 |
| Sep 20, 2026 | 13 |
| Sep 27, 2026 | 29 |
| Oct 4, 2026 | 27 |
| Oct 11, 2026 | 25 |
| Oct 18, 2026 | 12 |
| Oct 25, 2026 | 731 |
The Board ranks tools by four signals each week: new stars, new stars for the repo's size, added downloads and download growth beyond the market. How the ranking works.
Memvid alternatives
Other RAG, data and memory tools on the Board in the week of Oct 25–31, 2026, by momentum:
| Rank | Tool | Stars | Downloads/wk |
|---|---|---|---|
| #1 | XERJXERJ is a local-first search engine and MCP server for AI agents, packaged as one Rust binary with an Elasticsearch-compatible API. | 17.9k | – |
| #3 | LEANNLEANN is a Python vector database and RAG library that recomputes embeddings on demand to keep indexes small on personal devices. | 13.7k | – |
| #4 | GraphifyGraphify is an AI coding assistant skill that turns a codebase, with its docs, PDFs, images and videos, into a queryable knowledge graph. | 140k | 480.7k |
| #5 | MarkItDownMarkItDown is a lightweight Python utility from Microsoft that converts PDFs, Office documents, images, audio and other files to… | 193.2k | 4.7M |
| #6 | FirecrawlFirecrawl is an API and open source project to search, scrape, crawl, and interact with web pages, returning LLM-ready markdown or JSON… | 197.4k | 2.9M |
| #7 | Mem0Mem0 is a memory layer for AI agents and apps that stores and retrieves user, session, and agent context across interactions. | 68.2k | 884.1k |
Recent Memvid releases
Memvid FAQ
Is Memvid free and open source?
Yes. Memvid's source code is on GitHub (memvid/memvid) under the Apache 2.0 license, so you can use it for free.
How do I install Memvid?
Run npm install -g memvid-cli. The install section above has the other options from the project's README.
Is Memvid still maintained?
Activity has slowed: the memvid/memvid repository was last updated on Jul 14, 2026, and the latest release, v2.0.140 on GitHub, came out on May 27, 2026.
How popular is Memvid?
Memvid has 17.3k GitHub stars as of Nov 1, 2026. It gained 731 new stars in the week of Oct 25–31, 2026, against a 12-week average of 102 a week. That made it #2 among RAG, data and memory tools on the Devtoolwatch Board.
Memvid vs Mem0: how do they compare?
Memvid and Mem0 are both tracked by Devtoolwatch as RAG, data and memory tools. Memvid has 17.3k GitHub stars and Mem0 has 68.2k. Mem0 is a memory layer for AI agents and apps that stores and retrieves user, session, and agent context across interactions. Mem0 review.
What are the best alternatives to Memvid?
Other RAG, data and memory tools on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: XERJ (#1), LEANN (#3), Graphify (#4), MarkItDown (#5), Firecrawl (#6). The alternatives table above has their numbers.
What is Memvid?
Memvid is a memory layer for AI agents. It packages data, embeddings, search structures, and metadata into a single .mv2 file. Agents retrieve directly from that file instead of using a RAG pipeline or server-based vector database.
How does Memvid work?
Memvid writes memory as an append-only sequence of immutable Smart Frames, with design ideas borrowed from video encoding. The .mv2 file holds a header, an embedded WAL, compressed data segments, a Tantivy full-text index, an HNSW vector index, a time index, and a footer.
How is Memvid different from a typical RAG pipeline?
Memvid replaces the separate database and services of a typical RAG setup with one file. That file holds data, embeddings, and indexes. Memvid automatically chunks, embeds, and indexes imported content, and combines BM25 lexical search with vector search.
Does Memvid store video or use QR codes?
No. Memvid borrows ideas from video encoding to organize memory, but it does not store video. The QR-based Memvid v1 is deprecated.
Does Memvid work offline?
Yes. Memvid is described as working fully offline. It can use local ONNX embedding models, and the memory file can be deployed on-prem or air-gapped. OpenAI embeddings are optional.
Which embedding models does Memvid support?
Memvid supports local ONNX models: bge-small-en-v1.5 (default), bge-base-en-v1.5, nomic-embed-text-v1.5, and gte-large. Through OpenAI, it supports text-embedding-3-small, text-embedding-3-large, and text-embedding-ada-002.
Sources
Memvid README badge
Maintainers can show the tool's weekly Board rank in their README:
[](https://index.devtoolwatch.com/tools/memvid-memvid/)How this page is made: Devtoolwatch writes each review from the project's README, releases, package metadata and website, with an AI model, then checks every claim against the source it cites and drops what the sources don't support. The numbers come from Devtoolwatch's daily GitHub snapshots, GitHub's stargazer history, and the npm and PyPI download APIs. Spotted an error? Email [email protected].