Sub-2ms Pre-Filtered
Vector Search at Planet Scale.
MassEngine is a next-generation kNN vector database built from the ground up for multi-tenant AI & RAG systems. Eliminate selective-filter latency cliffs while cutting cluster memory footprints by 7×.
No Filter Cliffs. No Memory Bloat.
Evaluated on an identical 2-node replicated cluster (RF=2, 8 Shards, 2,000,000 × 1024-d English Wikipedia BAAI/bge-m3 embeddings, 10 concurrent clients). Select a filter selectivity below to compare throughput, latency, and memory:
Everything Modern RAG Needs in One Engine
Replace fragile multi-system pipelines with a single, deterministic search layer.
True Multi-Tenant Isolation
Host thousands of customer workspaces on shared infrastructure with zero noisy-neighbor degradation, scoped RBAC API keys, per-tenant QoS rate limits, and instant GDPR tenant purge.
High-Throughput Reads & Writes
Ingest over 72,000 vectors/sec with instant availability while sustaining up to 6,993 filtered QPS and sub-millisecond query latency under concurrent production load.
Rich Metadata Pre-Filtering
Filter on arbitrary JSON attributes using expressive operators ($eq, $ne, $gte, $lte, $in, $contains), date ranges, and category bitmaps with zero recall penalty.
In-Place Metadata Patching
Update document permissions, pricing, tags, or categories in place via PATCH without re-computing or re-uploading expensive high-dimensional embeddings.
Flexible Dimension Support
Native support for any embedding model dimension—from compact 128-d and 384-d models up to 1536-d and 3072-d frontier embeddings.
Serverless Cloud or BYOC
Start instantly on MassEngine Serverless, or deploy our zero-dependency static binary directly inside your own AWS, GCP, or bare-metal VPC with automated self-healing.
Clean REST & SDK Interface
Simple, predictable JSON APIs and typed SDKs for Go, Python, and TypeScript.
POST https://api.massengine.com/v1/search
Authorization: Bearer mk_live_9f8a7b6c...
Content-Type: application/json
{
"vector": [0.021, -0.044, 0.089, "... 1024 dims"],
"k": 10,
"nprobe": 16,
"rerank_pool": 64,
"min_score": 0.75,
"metadata_filter": {
"department": { "$in": ["security", "legal"] },
"clearance_level": { "$gte": 2 },
"tags": { "$contains": "approved" }
}
}
Query a Live 20,000,000-Vector Cluster
Run real-time kNN vector search and hardware pre-filtered queries against 20,000,000 English Wikipedia passages (1024-d embeddings across 8 shards, RF=2)—with zero data upload or code required.
RF=2) • Corpus: 20,000,000 × 1024-d Vectors
POST /v1/search)
{
"vector": [
0.0241,
-0.0188,
0.0412,
-0.0093,
"... (1024 float32 dims)"
],
"k": 8,
"nprobe": 16,
"rerank_pool": 64,
"metadata_filter": {
"category": { "$eq": "science" }
}
}
Customer Usage & Telemetry Portal
Every workspace includes built-in latency percentile histograms, partition-level byte metering, and real-time usage tracking (GET /v1/tenants/{tenant}/telemetry).
Search Latency Distribution (ms)
Live Histogram
Real-Time Usage Counters Per-Workspace Metering
| Scope / Partition | Routing Mode | Live Vectors | Logical Size | Status |
|---|---|---|---|---|
| demo-cloud / default | supercell (8 shards, RF=2) | 20,000,000 | 83.46 GiB | Healthy |
NVMe-Tiered Serverless, Pinned Hot-RAM Pods, or In-VPC BYOC
Start instantly with NVMe-backed Serverless Pay-As-You-Go (where idle indexes page to NVMe and you never pay for pinned DRAM), reserve Dedicated Hot-RAM Pods for guaranteed sub-2ms Stage-1 SLAs, or run inside your own cloud VPC.
Pay-As-You-Go
Disk-backed segments & cold metadata paged on demand from NVMe. Ideal for elastic workloads, dev/staging, and multi-million vector corpora with bursty traffic.
RF=2 included)
$0.35 / GiB-month
Pinned Hot-RAM Pods
Stage-1 compressed vector blocks pinned 100% in dedicated DRAM with NVMe Tier-2 reranking for ultra-low tail latency under sustained high QPS.
RF=2)
$0.35 / GiB-month
Bring Your Own Cloud (BYOC)
Deploy MassEngine's zero-dependency static binary directly inside your own AWS, GCP, or bare-metal VPC so sensitive vectors never leave your environment.
Request Demo & Early Access
Enter your work email and company to immediately unlock the live 8-shard Wikipedia cluster playground above and reserve your Private Beta credentials.