Everything Announced at MongoDB.local NYC 2026

MongoDB.local NYC 2026 is officially a wrap, and what an event it was!

An array of analysts, builders, businesspeople, caterers, engineers, fire marshals, media members, photographers, and trained elephants* chose to spend the day with us at Pier 36 in downtown New York, and we couldn’t be more thrilled. We just hope everyone had half as much fun as we did (it was a lot of fun).

The theme of this year’s event was how developers and enterprises can build anything and change everything with MongoDB. We released capabilities that make building AI applications and agents easier and faster for customers of all sizes—proving that MongoDB is the intelligent data platform for modern applications in the multi-cloud and AI era.

Specifically, the general availability of MongoDB 9.0, and the public previews of Atlas Infinite and Atlas Agent Engine represent the most significant expansion of our platform yet. Meanwhile, Voyage AI by MongoDB’s best-in-class retrieval capabilities—as measured by the Retrieval Embedding Benchmark (RTEB)—give agents and applications better accuracy at lower cost.

But those were hardly the only announcements we shared! Read on for details about everything launched at MongoDB.local NYC 2026.

MongoDB Atlas: The multi-cloud, intelligent data platform

Built for real-time operational data, best-in-class retrieval (with hybrid search and models from Voyage AI by MongoDB), and for running agents in production, MongoDB Atlas offers organizations a single platform to build, deploy, and manage AI applications and agents at any scale, across any cloud, on their terms.

Here’s everything we announced related to MongoDB Atlas:

• Atlas Infinite (Public Preview)
• Enterprise workloads are getting bigger, more variable, and harder to predict, especially as agentic applications introduce unpredictable demand. Customers need a data layer built for that reality, not one that's stretched to accommodate it. Atlas Infinite separates compute from storage so each can scale independently, giving every workload the elasticity, responsiveness, and retrieval accuracy necessary.

• Atlas Agent Engine (Public Preview)

• MongoDB Atlas is the intelligent data platform where agents act on real-time data, and Atlas Agent Engine is how you build, deploy, and govern those agents. Built-in memory and retrieval means agents remember, act on accurate context, and use tokens more efficiently. A single control plane governs every agent action with a verifiable identity, giving enterprise-grade oversight. It runs on the MongoDB Atlas infrastructure that leading enterprises trust, so your agents work on the data you already have, right where it lives.

• Atlas Ephemeral Clusters (Public Preview)

• Atlas Ephemeral Clusters allow AI agents and developers to spin up temporary MongoDB Atlas databases, removing human-in-the-loop friction to accelerate building and prototyping applications. With Ephemeral Clusters, agents can create a database in seconds and immediately receive credentials and a connection string—allowing them to build without delay.

• Atlas SQL Dedicated Cluster (GA)

• Now generally available, Atlas SQL Dedicated Cluster gives your BI tools direct SQL access to your dedicated cluster, with predictable pricing tied to what you already spend on MongoDB Atlas.

• New Metrics UI in Atlas (GA)

• The new Metrics UI gives MongoDB Atlas operators a clear place to understand cluster health, reduce navigation overhead, and investigate issues faster and more confidently. 

MongoDB 9.0: The best version of MongoDB, ever

MongoDB 9.0 is the best-performing version of the world's most popular document database. MongoDB 9.0 strengthens the foundation customers already run on, with major gains in performance, security, and scalability.

Here’s everything we announced related to MongoDB 9.0:

• MongoDB 9.0 (GA)
• Now generally available, MongoDB 9.0 is the best version of MongoDB ever built, delivering up to 2x higher throughput along with major advances across performance, scalability, observability, security, and builder productivity. It’s a core pillar of MongoDB’s intelligent data platform for AI, raising the ceiling for the most demanding enterprise and AI-native workloads. MongoDB 9.0 also extends our industry-first Queryable Encryption and private networking so even the most regulated organizations can build on it.

• Intelligent Workload Management (GA)

• MongoDB 9.0 features Intelligent Workload Management (IWM), a managed availability capability that keeps MongoDB Atlas clusters responsive and predictable even under overload. Only available via MongoDB Atlas, IWM queues transient spikes, keeps short-running operations moving and long-running operations stable, and protects cluster health with no application changes required. For sustained overload, customers can optionally enable Load Shedding to actively reject excess work. IWM is activated by default across all dedicated MongoDB 9.0 Atlas clusters at no extra charge.

• Queryable Encryption prefix and suffix query support (GA)

• A groundbreaking, industry-first technology, MongoDB Queryable Encryption enables customers to encrypt sensitive data—such as personally identifiable information—while running expressive queries directly on the data without requiring any decryption. Queryable Encryption now supports prefix, suffix, and substring searches directly on fully encrypted fields, so sensitive data stays protected throughout its full lifecycle without slowing applications down.

• MongoDB Extension for Hibernate ORM (GA)

• The MongoDB Extension for Hibernate ORM gives Java developers the fastest path to building production-ready applications on MongoDB using the familiar Hibernate ORM programming model, the de facto standard ORM framework for many Java enterprise applications. Organizations can use existing Hibernate skills and patterns with MongoDB, making it easier to modernize applications, accelerate development, and reduce migration complexity.

Best-in-class search and retrieval

Every answer, decision, and action an AI agent takes depends on retrieving accurate, relevant information. Better retrieval helps agents understand enterprise data more accurately, consume fewer tokens, and act more confidently in production.

MongoDB is dedicated to empowering customers to build retrieval systems that agents can rely on. Today’s announcements build on that foundation, further improving retrieval accuracy, reducing operational overhead, and giving teams more control over where their content is processed.

Here’s everything we announced related to search and retrieval:

• Voyage rerank-3 and rerank-3-lite (GA)
• rerank-3 is a pointwise reranker designed as a drop-in upgrade to rerank-2.5, offering improved performance across nearly all domains with a specific focus on long-context and code capabilities. rerank-3-lite matches the retrieval quality of rerank-2.5 at 40% of the price.

• Europe Data Residency for Atlas Embedding and Reranking (GA)

• MongoDB Atlas customers can use Voyage AI’s embedding and reranking models for AI-powered search while relying on MongoDB to process submitted content within the Europe Geography.

• Native Reranking (GA)

• MongoDB now offers reranking functionality directly in the aggregation pipeline using Voyage AI’s best-in-class reranking models.

• Accelerated Search Aggregations (Public Preview)

• MongoDB is upgrading Atlas Search's facet feature so customers can facet on multiple fields at once and get more than just counts—stats like min, max, sum, and average—all in a single query, using the same data already indexed for search.

From thinking (and prototyping) to acting at scale

The announcements we shared at MongoDB.local NYC 2026 help customers move from prototype to production quickly.

They also build on an already-rock-solid foundation—our platform serves more than three trillion queries a day, and ~75% of the Fortune 100 rely on MongoDB for critical applications. Together, the launches of MongoDB 9.0, Atlas Infinite, and Atlas Agent Engine (+ Voyage AI’s industry-leading models) make MongoDB the intelligent data platform for modern applications in the multi-cloud and AI era. 

*Note that there were no trained elephants at MongoDB.local NYC 2026, sadly. Next year!?