Neon
Neon is a serverless Postgres platform built to help developers ship faster with a fully managed, autoscaling database that “just works.” It blends th...
Rate this app
Neon Overview
Key Features & Capabilities
Neon automatically adjusts compute based on demand and can scale to zero when idle to reduce costs. Paid tiers allow you to control or disable scale-to-zero for predictable latency under load.
Instantly branch data and schema with copy-on-write storage for safe experimentation, testing, and preview environments. Branches only incur storage for deltas, enabling efficient multi-env workflows.
Offload read-heavy workloads to replicas that scale down to zero when idle. This boosts performance while avoiding additional storage costs for replicas.
Recover to any point in time with configurable PITR windows (up to 30 days on higher tiers). Billing is based on actual data changes during the selected window, not per-restore events.
Query Postgres over HTTP with the serverless driver and a PostgREST-compatible Data API, ideal for serverless and edge runtimes. Combine with Neon Auth and RLS for secure, frontend-friendly access.
Pricing Plans
Starter/Free
- Up to 20 projects and 10 branches per project included
- 100 CU-hours per project (compute cap)
- Sizes up to 2 vCPU, 8 GB RAM
- Autoscaling with scale-to-zero after 5 minutes of inactivity
- 0.5 GB storage per project included
- Instant Restore up to 6 hours or 1 GB of data changes
- 1-day monitoring retention
- Read replicas and built-in high availability
Pro/Plus/Core (Launch)
- 100 projects; 10 branches per project included
- Compute: $0.106 per CU-hour
- Sizes up to 16 vCPU, 64 GB RAM
- Autoscaling (scale-to-zero after 5 minutes; can be disabled)
- Storage: $0.35 per GB-month
- Instant Restore window up to 7 days ($0.20 per GB-month based on changed data)
- Public egress: 100 GB included then $0.10/GB
- 3-day monitoring retention
- All Free features plus longer PITR and higher limits
Teams/Business (Scale)
- 1,000+ projects; 25 branches per project included
- Compute: $0.222 per CU-hour
- Sizes up to 56 vCPU, 224 GB RAM
- Autoscaling configurable
- Storage: $0.35 per GB-month
- Instant Restore window up to 30 days ($0.20 per GB-month based on changed data)
- Public egress: 100 GB included then $0.10/GB; Private network transfer $0.01/GB
- Metrics/logs export; 14-day monitoring retention
- Private Networking / IP allow rules; 99.95% SLA
- SOC 2 report access; HIPAA available
- All Launch features plus enterprise-grade networking, observability, and compliance
Enterprise
- All Scale features
- Custom resource limits and quotas
- Private networking and IP allow rules
- Dedicated support and onboarding
- Custom SLAs and compliance assistance (SOC 2, HIPAA)
- Security reviews and advanced governance options
Paid plans have a $5 monthly minimum. Additional branches beyond included quotas are billed at ~$0.002 per branch-hour (prorated). Instant Restores (PITR) are billed at $0.20 per GB-month based on data changes during the selected window. Public egress on paid plans includes 100 GB/month then $0.10/GB; Free includes 5 GB. Scale to zero occurs after 5 minutes of inactivity by default (configurable/disable on paid plans). All databases include autoscaling, read replicas, high availability, monitoring, connection pooling (pgBouncer), and a rich Postgres extensions library. Startup credits are available via the Startup Program. An Agent Plan exists for platforms provisioning thousands of databases.
Pros & Cons
Pros
- Instant branching with copy-on-write enables safe CI/CD, previews, and rapid schema iteration without full copies.
- Autoscaling with scale-to-zero reduces idle costs while still handling bursts, ideal for serverless and spiky workloads.
- Read replicas that scale down to zero and avoid extra storage provide cost-efficient performance for read-heavy apps.
- Edge-ready HTTP serverless driver and PostgREST-compatible Data API simplify frontend/serverless access to Postgres.
- Enterprise-grade options on higher tiers: private networking, 99.95% SLA, metrics/logs export, SOC 2 and HIPAA availability.
Cons
- ×Usage-based pricing with a monthly minimum and egress overages can make costs unpredictable under heavy or spiky load.
- ×Free tier limits (100 CU-hours, 0.5 GB storage, small instance sizes) are restrictive for production workloads.
- ×Scale-to-zero can introduce cold-start latency after periods of inactivity unless disabled on paid plans.
- ×Longer PITR windows, private networking, metrics/log export, and extended monitoring retention require higher-tier plans.
- ×Neon Data API is marked as beta, which may entail evolving behavior and features.
User Reviews
Alex Klarfeld – CEO and co-founder, Supergood.ai
Neon allows us to develop much faster than we’ve even been used to
Edouard Bonlieu – Co-founder, Koyeb
Neon's serverless philosophy is aligned with our vision: no infrastructure to manage, no servers to provision, no database cluster to maintain
Léonard Henriquez – Co-founder and CTO, Topo.io
The killer feature that convinced us to use Neon was branching: it keeps our engineering velocity high
Himanshu Bhandoh – Software Engineer, Retool
We've been able to automate virtually all database tasks via the Neon API, saving us a tremendous amount of time and engineering effort
Frequently Asked Questions
Is there a free plan or free trial?
Yes. Neon offers a Free plan ($0, no card required) with up to 20 projects, 10 branches per project, 100 CU-hours per project, 0.5 GB storage, and sizes up to 2 vCPU/8 GB. It includes autoscaling, read replicas, and built-in high availability.
What is a project in Neon?
A project is the top-level container for your database environment, including the primary database, its branches, and compute resources—similar to a GitHub repo with multiple branches.
What is a CU and how is compute billed?
A CU (Compute Unit) equals 1 vCPU + 4 GB RAM. Compute is billed by CU-hours based on the size your compute runs at and the time it runs (e.g., 2 CU for 3 hours = 6 CU-hours).
How is storage billed?
Storage is billed on actual usage in GB-months (metered hourly and summed monthly). You pay for what you store, not the maximum size reached.
How do branches affect storage?
Branches use copy-on-write. Child branches do not add storage unless they diverge; only the delta vs. root counts toward your storage bill.
How are Instant Restores (PITR) billed and what are the limits?
Billing is based on data changes during the selected window at $0.20/GB-month. Free includes up to 6 hours or 1 GB of changes. Launch allows 0–7 days; Scale/Enterprise up to 30 days.
Why do Launch and Scale have the same monthly minimum?
Premium Scale features are priced via a higher per–CU-hour compute rate rather than a higher flat fee, so you only pay more when you use the additional capabilities.
How can I control costs?
Set autoscaling maximums per branch to cap CU usage and disable scale-to-zero on paid plans to control latency. Monitor egress and branch counts; extra branches are billed per branch-hour.
Do you offer startup credits?
Yes. Venture-backed early startups can apply to the Neon Startup Program for credits.
Start for Free
Join thousands of developers who are already using Neon to enhance their workflow and productivity.
More Database & Storage
Pinecone
Pinecone is a fully managed, serverless vector database built to power high‑quality retrieval for AI applications at production scale.
Convex
Convex is a full cloud backend designed for modern TypeScript/JavaScript applications that need realtime data, strong transactions, and an integrated developer experience.
Supabase
Supabase is a Postgres-first backend platform that lets developers build modern applications fast without managing infrastructure.
Firebase
Firebase is Google’s app development platform that lets teams build, ship, and scale web and mobile applications faster with a unified backend and operations toolset.
Render
Render is a modern cloud platform that lets teams build, deploy, and scale full‑stack applications with minimal DevOps overhead.