Cinch is in public alpha.

// PRICING

Pay for what you use.

Three tiers of storage. Active data is fast. Idle data is cheap.

HOT
$10
GB-mo · RAM
WARM
$0.50
GB-mo · NVMe
COLD
$0.20
GB-mo · cloud

Plans

STARTER
$10/mo
+ usage · $10 included
  • Unlimited databases
  • All database types
  • Auto-stop and auto-wake
  • Read replicas
PRO
$50/mo
+ usage · $50 included
  • Everything in Starter
  • Priority support
  • Audit logging
  • Team access
COMING SOON
ENTERPRISE
Custom
For agent infra at scale
  • Everything in Pro
  • Volume discounts
  • SLA guarantees
  • Dedicated support

All plans use the same per-GB pricing. The plan minimum is your included usage — go over and you just pay the difference. No overages, no surprises.

What's included

Unlimited databases
All database types (cache, graph, SQL)
Auto-stop and auto-wake
Cloud archival for idle databases
HTTP API + SDK access
Per-database isolation
Read replicas (same price)
No per-request fees

Estimate your cost

Drag the sliders. See what you'd pay.

This calculator is a simulation — actual costs will vary based on your real usage patterns.

DATABASES50
AVG SIZE PER DATABASE500 MB
ACTIVE AT ANY TIME5%
HOT DATA (IN RAM)20%
377 MB in RAM · 7.8 GB on NVMe · 16.6 GB in cloud storage
Power-law distribution · 10-run average
HOT (RAM)$3.68/mo
WARM (NVMe)$3.91/mo
COLD (CLOUD)$3.32/mo
TOTAL
$10.91/mo

How pricing works

Every database has data distributed across three tiers. You configure each database: how much data stays hot in RAM vs. on disk, and how quickly idle databases archive to cloud storage.

  • Active database: Pays $0.50/GB for all data on disk, plus $10/GB for the fraction you configure to stay hot in RAM. You control the split.
  • Idle database: After your configured idle period, the database archives to cloud storage at $0.20/GB. No server running. No disk allocated.
  • Waking up: First request to an archived database restores it from cloud storage to NVMe. Subsequent requests are full speed.
  • Read replicas: Same price as the primary. No markup for replicas.

Example costs

ScenarioDataTraditionalCinch
Single active cache (1GB)1GB on disk, ~100MB hot~$15/mo (ElastiCache)~$1.50/mo
10 agent caches (500MB each)2 active, 8 idle~$75/mo (10 instances)~$1.80/mo
100 tenant databases (1GB each)20 active, 80 archived~$1,500/mo~$26/mo
1,000 tenant databases (1GB each)100 active, 900 archived~$15,000/mo~$230/mo
Multi-region app (5GB, 3 regions)3 replicas, all active, 20% hot~$150/mo (3x RDS)~$38/mo
Seasonal SaaS (500 tenants, 1GB)50 active, 450 archived~$7,500/mo~$165/mo
IoT fleet (100K devices, 1MB each)1% active, 99% archived~$500/mo (DynamoDB)~$21/mo
Agent infra platform (50K stores, 100MB)2% active, 50% hot, rest archived~$20,000/mo~$1,500/mo

Traditional prices based on equivalent managed services (ElastiCache, Neo4j Aura, RDS). Cinch prices assume ~10% of data hot in RAM for active databases.

No surprises

No per-request fees. No bandwidth charges. No connection limits. No minimum commitment. You pay for storage across three tiers. That's it.

FAQ

Why no free tier?

Every database we run has real infrastructure costs, and we'd rather give alpha users great service than spread thin with a free tier. Sign up and start building.

What happens if I go over my plan's included usage?

You just pay the per-GB difference. No overage penalties, no surprise multipliers. Same rates as shown above.

How fast does a cold database wake up?

Milliseconds. The first request triggers a restore from cloud storage to NVMe. Subsequent requests hit full speed. No manual intervention needed.

Try the public alpha.

Sign up with GitHub or email. Free to start.

GET STARTED →