Neon Serverless Postgres

Overview

Neon is a serverless PostgreSQL platform that combines traditional database capabilities with modern cloud-native features, offering native pgvector support for AI and vector search applications.

Key Features

Serverless Architecture

• Compute-Storage Separation: Independent scaling of compute and storage
• Instant Provisioning: Database creation in under 1 second
• Auto-Scaling: Automatic compute scaling based on demand
• Scale to Zero: Databases automatically pause when idle
• Pay-per-Use: Only pay for active compute time

pgvector Support

• Native pgvector extension support
• HNSW indexing for fast similarity search
• IVFFlat indexes available
• Full PostgreSQL 15+ compatibility
• Embedding dimensions up to 2000

Branching

• Database branching similar to Git
• Instant branch creation from any point in time
• Perfect for development/staging workflows
• Zero-cost copies for testing

Security & Multi-Tenancy

• Row Level Security (RLS)
• SOC 2 compliant
• Point-in-time recovery

Use Cases

• Multi-tenant AI apps (RAG, semantic search)
• Serverless AI inference pipelines
• Global development teams with branching

Comparisons

vs Dedicated Vector DBs: Full Postgres SQL (joins, analytics, ACID) + vectors vs specialized but limited querying.

Feature	Neon + pgvector	Dedicated (e.g., Pinecone, Weaviate)
Serverless Scaling	Instant, scale-to-zero	Usage-based
SQL + Vectors	Native hybrid	Often separate
Branching	Git-like	None
Cost for Idle	$0	Minimums

Pricing

Free Tier

• 512 MB storage
• 0.5 compute hours/month
• Unlimited projects

Paid Plans

• Launch: $19/month base + usage
• Scale: Custom
• Enterprise: Volume discounts

Compute per active hour, storage per GB-month.