MongoDB has released an open weight model (nano) in the voyage-4 series of embedding models. It is on huggingface (see follow-on post). Apache 2. This is an excellent way to quickly do a prototype of RAG type things or AI style vector search.
Note you do not need to use MongoDB at all to use this. (Although I would. It could be a free version like the community edition locally or Atlas free tier in the cloud.)
The era of embedding models is evolving. 🚀
With voyage-4-large, we’ve moved to Mixture of Experts (MoE) to shatter the scaling ceiling.
The results: ✅ Massive drop in inference cost and latency ✅ New frontier for retrieval accuracy.
Curious about how we implement MoE embeddings?
Read the full technical breakdown of how we optimized design choices to push the Pareto frontier: https://t.co/HmfYoCoUxH
📢 ICYMI: We recently released the Voyage 4 model family.
Voyage 4 features an industry-first shared embedding space, eliminating the need to re-index your data when switching between models in the series.
This breakthrough is designed to solve the trade-off between frontier-level accuracy and production efficiency.
Forget genres - what if you could search for a movie like this:
“Something uplifting after a rough day”
“A film that’ll make me cry”
In this new tutorial with @huggingface, Arek Borucki takes you through how to build a mood-based recommendation engine using voyage-4-nano and MongoDB Atlas Vector Search.
Give it a try! https://t.co/cCwX8jZphe
The Embedding and Reranking API on MongoDB Atlas is officially in Public Preview!
@VoyageAI frontier embedding and reranking models, available as a simple standalone API. Use it with any stack, or combine it with Atlas Vector Search.
What you get:
• Access the new Voyage 4 series for better context and reduced hallucinations.
• Your data, vector search, and models all under one control plane.
• Simple token-based pricing + 200M free tokens to start building.
Build end-to-end AI retrieval in Atlas, from data storage and vector search to embeddings and reranking
Read the full announcement: https://t.co/kx2VSFH5bO
Announcing our 2025 winners of MongoDB’s prestigious William Zola Award for Community Excellence: Arek Borucki and Kai Yong Lai!
Each year, this award recognizes an outstanding community member who embodies the legacy of William Zola, a Lead Technical Services Engineer at MongoDB who passed away in 2014.
Learn more about how our award winners have made an impact on the MongoDB community: https://t.co/oNEVg7wxOq
Behind every Vogue article and GQ story is a massive data challenge.
Go inside @CondeNast’s move to MongoDB, and how it changed the way they deliver content at a global scale. https://t.co/630539gRvC
voyage-3.5 and voyage-3.5-lite are available today! The first 200 million tokens are free. Get all the details in our latest blog: https://t.co/3lmVCNK7OT
We've been building the next leap in retrieval accuracy and efficiency.
Tomorrow, we make it available for everyone.
See you here at 10:00 am PT for the announcements 🚀
Watch the livestream: https://t.co/pfGg3hjkyo
For the fourth consecutive year, MongoDB has been named a Leader in the @Gartner_inc® Magic Quadrant™ for Cloud Database Management Systems (CDBMS)!
To learn more about this recognition and how we believe it reflects our commitment to customer innovation, read the full report here: https://t.co/6XINOdJ7zb
MongoDB’s B+tree was engineered lock-free for both reads and writes. Results:
⚡ 50% throughput improvement for reads
⚡ 4x gains for high-contention updates
I have some news to share. After 11 incredible years, I’ve decided to retire as CEO of MongoDB. This was not an easy decision, but it’s the right time for a thoughtful transition.
Ever wish you could search encrypted data without decrypting it first? 🔐
At MongoDB.local London, we chatted with Joel Odom, Staff Product Manager, Security at MongoDB about how Queryable Encryption makes it possible.
Learn more about Queryable Encryption 👉 https://t.co/P0oFqWinRV
To evaluate embeddings and retrieval, we need more benchmarks beyond MTEB that are less vulnerable to overfitting. That’s why RTEB was just beta-launched!
⚖️ Both open and held-out datasets to prevent overfitting to evaluation sets.
🌍 Realistic datasets from critical enterprise domains like law, healthcare, code, and finance.
🔎 Only focus on retrieval applications with relevant large-scale datasets.
Check out the blog and leaderboard on @huggingface and join the community in building a stronger, more reliable benchmark.
Blog: https://t.co/T6ZcROuVCg
Queryable Encryption with prefix, suffix, and substring query support is now in Public Preview! 🔐
With MongoDB 8.2, we’ve expanded our industry-first encryption-in-use technology. Now, in addition to equality and range queries, you can securely run prefix, suffix, and substring queries directly on encrypted data.
Why this matters:
• Keeps sensitive data encrypted at rest, in transit, and in use
• Eliminates risky workarounds like leaving fields unencrypted or using external indexes
• Simplifies compliance with regulations like GDPR, HIPAA, and PCI-DSS
• Available at no extra cost across Atlas, Enterprise Advanced, and Community Edition
Listen to the #MongoDBlocal keynote announcement with Osmar Olivo, Senior Director, Database Product Management, or learn more on our blog: https://t.co/M66jfslOm7
We’re excited to announce the integration of MongoDB Atlas with the MCP Toolbox for Databases!
MongoDB Atlas + @googlecloud's MCP Toolbox give developers the speed, flexibility, and scalability to build the next wave of AI-powered applications.
👉 Learn how to get started: https://t.co/6ViwPs4t7e