Last updated

Deployment Options

Papr Memory is available in two deployment modes to suit different needs, from rapid prototyping to enterprise production deployments.

Ranked #1 on Stanford's STaRK benchmark with 91%+ accuracy and <100ms retrieval. Run it open-source self-hosted, or use our managed cloud with predictive memory built in—same API, full control.

Choose Your Deployment

☁️ Papr Cloud (Managed)

Fully managed, production-ready service

  • 5-minute setup - API key and you're done
  • Predictive memory - <100ms retrieval with anticipatory caching
  • 91%+ accuracy - #1 on Stanford's STaRK benchmark
  • Zero infrastructure - We handle scaling, updates, backups
  • Gets better with scale - More data improves predictions
  • 24/7 support - Direct access to our engineering team

Best for: Production apps, fast deployment, teams focused on building features

Get Started with Cloud → | View Pricing →

🔒 Hybrid Cloud (Enterprise)

Managed service in your cloud

  • Your infrastructure - Deployed in your AWS/Azure/GCP account
  • We manage it - Zero operational burden, automatic updates
  • Full predictive features - All cloud capabilities included
  • Data sovereignty - Data never leaves your environment
  • Compliance ready - Meet regulatory requirements
  • Dedicated support - Enterprise SLA with dedicated team

Best for: Enterprises with data residency requirements, regulated industries, large-scale deployments

Talk to Sales → | Learn More →

🐳 Self-Hosted (Open Source)

Run on your own infrastructure

  • Same core engine - Vector + graph + predictive layer
  • Full control - Complete data sovereignty
  • On-premise - Keep data in your VPC/data center
  • Customizable - Modify and extend as needed
  • AGPL-3.0 - Open source license, transparent by design
  • Community support - GitHub discussions + Discord

Best for: On-premise requirements, maximum control, development/testing

Self-Hosted Setup Guide → | GitHub Repository →

Deployment Comparison

FeaturePapr CloudHybrid CloudSelf-Hosted
Setup Time5 minutes1-2 weeks30-60 minutes
Infrastructure ManagementFully managedFully managedYou manage
ScalingAutomaticAutomaticManual configuration
Updates & PatchesAutomaticAutomaticManual updates
Data LocationUS/EU regionsYour cloud accountYour infrastructure
SupportPriority supportDedicated + SLACommunity + GitHub
ComplianceGDPR readyYour compliance frameworkYour responsibility
Predictive Features✅ Advanced✅ Advanced✅ Basic
Custom Domains✅ Yes✅ Yes✅ Yes
API Compatibility100%100%100%
Cost ModelUsage-basedCustom contractInfrastructure + AI APIs

Feature Parity

Both deployments share the same core functionality with some operational differences:

FeaturePapr CloudSelf-HostedNotes
Core Memory APIIdentical API
Vector SearchSame algorithms
Knowledge GraphsSame graph engine
Custom SchemasFull schema support
GraphQL QueriesComplete parity
Document ProcessingPDF, Word, Text
Predictive Caching✅ Advanced✅ BasicCloud uses ML-optimized caching
Developer Dashboard✅ Built-in❌ Not includedCloud-only web UI
Monitoring Dashboard✅ Built-in⚙️ DIYSelf-hosted: Use Grafana/Prometheus
Multi-region Deployment⚙️ Your setupCloud: US-East, US-West, EU
SSO Integration✅ Enterprise⚙️ ConfigureSelf-hosted: Use Parse or custom
Automatic Backups⚙️ ConfigureSelf-hosted: Setup backup scripts
High Availability⚙️ ConfigureSelf-hosted: Setup replication
Rate Limiting✅ Built-in✅ ConfigurableBoth support
API Analytics✅ Dashboard⚙️ SetupSelf-hosted: Use logs + analytics

Legend:

  • ✅ = Included out of the box
  • ⚙️ = Requires configuration/setup
  • ❌ = Not available

Quick Decision Guide

Choose Papr Cloud if you:

  • Want to focus on building your app, not managing infrastructure
  • Need to get to production quickly (5-minute setup)
  • Prefer predictable pricing with usage-based billing
  • Want automatic scaling and performance optimization
  • Don't have strict data residency requirements

Choose Hybrid Cloud if you:

  • Need data to stay in your cloud environment (AWS/Azure/GCP)
  • Want managed service without operational burden
  • Have compliance requirements (HIPAA, SOC 2, FedRAMP, etc.)
  • Need predictive features with data sovereignty
  • Want enterprise SLA and dedicated support
  • Are in regulated industries (healthcare, finance, government)

Choose Self-Hosted if you:

  • Have strict on-premise or air-gapped requirements
  • Want complete control over infrastructure and customization
  • Have in-house DevOps expertise for managing services
  • Prefer to manage your own infrastructure costs
  • Need to audit or modify the source code
  • Want to contribute to the open-source project

Architecture Overview

Both deployment options share the same core architecture:

Intelligence Layer
Storage Layer
Application Layer
OpenAI API
Groq API
Deep Infra API
MongoDB
Metadata
Qdrant
Vectors
Neo4j
Graph
Redis
Cache
FastAPI Server

Key Difference: With Papr Cloud, all infrastructure components are fully managed. With Self-Hosted, you run and manage all components yourself.

API Compatibility

Both deployment options use identical APIs. Code written for one deployment works with the other:

# Same code works for both Cloud and Self-Hosted
from papr_memory import Papr

# Cloud deployment
client = Papr(
    x_api_key=os.environ.get("PAPR_MEMORY_API_KEY"),
    base_url="https://memory.papr.ai"  # Cloud URL
)

# Self-hosted deployment
client = Papr(
    x_api_key=os.environ.get("PAPR_MEMORY_API_KEY"),
    base_url="http://localhost:5001"  # Your self-hosted URL
)

# Same API calls work for both
memory = client.memory.add(
    content="This works the same everywhere",
    type="text"
)

Migration Between Deployments

Need to migrate between deployment options? Contact our team for assistance with data migration and transition planning.

Next Steps

Ready to get started?