Dedicated public ports on a static IP, tunneled directly to your machine. Built for vast.ai hosts, self-hosted services, homelabs, and anyone who needs a machine reachable from the internet.
Works on Ubuntu and other Linux hosts, including Raspberry Pi and ARM64 systems.
Start with a 1-day free trialYour ISP puts you behind carrier-grade NAT. Your machine has no public IP. Platforms like vast.ai, remote admin tools, and self-hosted services all depend on stable inbound reachability.
No public ports means broken inbound access.
Real ports on a real IP. Not HTTP tunnels, not dynamic URLs, and not single-port VPN hacks.
Your ports never change. A consistent public address that vast.ai and other platforms can rely on. No rotating IPs, no DDNS.
Paste a single command on your GPU rig. The installer handles everything: binary, config, systemd service. Live in under 60 seconds.
All inbound traffic is routed through TLS-encrypted tunnels to your machine. Your data stays private in transit.
Runs as a system service. If your connection drops or your machine reboots, the tunnel reconnects automatically. Zero babysitting.
NeedPorts monitors tunnel-server health and can move affected customers to healthy capacity when infrastructure degrades. Your install stays the same.
Forward any protocol. SSH, HTTP, custom AI inference endpoints, anything that listens on a port.
Each machine gets its own dedicated port range and authentication token. No one else can access or interfere with your ports.
No router configuration. No firewall rules. No networking knowledge required.
Choose how many ports and bandwidth you need. We generate a one-time setup token for your machine.
Paste the one-liner on your GPU rig. It installs the tunnel client, writes the config, and starts the service.
Your ports are immediately reachable from anywhere in the world. Point vast.ai at your new IP and start earning.
Start with a 1-day free trial on eligible plans. A valid payment method is required, then choose the plan that fits your machine or service footprint.
GPU hosting is just the start. We're testing and adding support for these use cases next.
Minecraft, Valheim, CS2 from home
Coming SoonSelf-host Nextcloud, Plex, and more
Coming SoonSSH/RDP into machines without public IPs
Coming SoonAccess devices remotely without VPN
Coming SoonHost WireGuard or OpenVPN servers
Coming SoonMost alternatives break down because they do not give you a stable public endpoint with enough dedicated ports.
| Option | Static public endpoint | Many dedicated ports | Easy setup | Good fit for GPU hosts |
|---|---|---|---|---|
| NeedPorts | Yes | Yes | Yes | Yes |
| Typical VPN port forwarding | Sometimes | No, usually one port | Usually easy | Usually not ideal |
| Asking ISP for a public IP | Sometimes | Maybe | Often slow or unavailable | Can work if offered |
| Running your own VPS relay | Yes | Yes | No, more operational work | Can work, but higher effort |
Pick the rough size of your setup and we will suggest a starting point.
The short version: if your ISP or hosting setup puts you behind CGNAT, NeedPorts gives you a stable public endpoint and dedicated forwarded ports without making you rework your network.
CGNAT means your ISP places your machine behind a shared public IP, so you cannot open inbound ports directly. Platforms like vast.ai need reachable inbound ports to route workloads to your host. Without public reachability, your machine may not be listable or usable.
Most VPN providers give you one forwarded port at best, and often rotate or restrict what you can expose. NeedPorts gives you a dedicated static public endpoint with a reserved block of ports for your machine, which is a better fit for GPU hosts and multi-service setups.
Yes. Your machine connects to the NeedPorts tunnel service and receives a dedicated forwarded port range on a stable public IP. That makes it suitable for vast.ai, SSH, web services, inference endpoints, and other workloads that need consistent inbound access.
Usually about a minute. After payment, NeedPorts provisions your port range automatically and gives you a one-line install command. Run it on your machine and the client installs, writes the config, starts the service, and connects.
You are redirected to a success page that shows your install command once provisioning completes. Run that command on the target machine and NeedPorts will configure the tunnel client with your assigned port range automatically.
Yes. The service is designed for real port forwarding, not just HTTP tunnels, so it can support TCP and UDP workloads that need direct inbound connectivity.
Yes. Monthly plans can be cancelled anytime. Annual plans are billed once per year at the discounted rate. Subscription management is handled through Stripe.
No. vast.ai is the primary use case today, but NeedPorts also works for any machine behind CGNAT that needs a stable public IP and dedicated forwarded ports for remote access or hosted services.
If you need stable remote access to a home server, homelab, dashboard, API, or self-hosted service behind CGNAT, start with these pages.
Reach SSH, dashboards, and services remotely when your ISP uses CGNAT.
GuideWhat to do when remote access fails because your connection has no usable public IP.
GuideHow to get a stable public endpoint without paying your ISP for a special plan.
GuideExpose SSH, APIs, dashboards, and other services when normal port forwarding fails.
GuideGet stable public ports when cellular internet or fixed wireless is behind CGNAT.
GuideKeep inbound access working when your backup WAN takes over.
GuideUse one public endpoint across primary and failover internet paths.
GuideIf you're comparing options or diagnosing why your host or service is unreachable, start here, or browse the full guides hub.
How NeedPorts helps vast.ai hosts behind CGNAT.
GuideWhy CGNAT breaks inbound connectivity and how to work around it.
GuideDiagnose GPU hosts that appear online but time out from the public internet.
GuideSee how outbound tunnels create stable public inbound reachability.
ExplainerCreated by Vast hosts for Vast hosts, with the same reachability problems in mind.
AboutSee the full guide hub for GPU hosting, homelab, self-hosting, and networking explainers.
HubNo. vast.ai is one strong use case, but NeedPorts also supports self-hosted and homelab setups. The live installer now supports an explicit self-hosted mode that skips vast-specific integration and includes a simple verification helper.
Start with a 1-day free trial on eligible plans. Choose your plan, use case, and billing interval, then continue securely through Stripe with a valid payment method. Annual plans are billed once per year at a 50% discount. Your ports are provisioned automatically and you'll get an install command immediately after checkout.
A name for your machine. Alphanumeric, hyphens, underscores. Max 32 chars.
1-day free trial for eligible new customers. Valid payment method required. Monthly plans cancel anytime.