What's new in VorticPanel.
Every release and what changed in it. The newest is 0.3.72, from October 24, 2026.
October 2026
Versions 0.3.0 to 0.3.72
- Imports from VirtFusion and Virtualizor: pick the servers, copy them while they run or adopt them in place, and roll back for up to 14 days.
- Live moves between nodes, and moves with one restart between Intel and AMD or between Linux distributions.
- Resellers sell servers and GPU instances within the quotas you set.
- GPU instances in containers, with JupyterLab over HTTPS and the customer's own Docker image where you allow it.
- Billing through WHMCS or Blesta, with one-click sign-in for customers.
- Private networks between nodes encrypted with WireGuard, and nodes that only take orders signed by your panel.
- IPv6 pools of any size, and backups made while servers keep running.
Show each releaseHide each release
- A node's SSH can be on any port: the panel reads it from the node and shows it on the node's Overview, and moves to the node connect there.
- Imports read each source hypervisor's SSH port from the source panel when it gives one, and Change port beside the hypervisor sets it.
- GPU instances show their usage over the last hour, day, week or month: GPU load, card memory, temperature, power, CPU, memory and network.
- An instance can be renamed from its Details too.
- GPU instance pages show the two disks plainly: Your files (/workspace, kept) and System disk (reset when started afresh).
- CUDA works in GPU instances after their host restarts; hosts missing the device CUDA needs are repaired when the update reaches them.
- Own GPU images from private registries pull with the login given.
- Own GPU images take an image's page address from Docker Hub, GitHub, Quay or NVIDIA NGC, with or without https://, and show which image they'll run.
- When an image can't be pulled, the message says whether the registry has no such image, is limiting pulls or can't be reached.
- GPU instance pages show the two disks plainly: Your files (/workspace, kept) and System disk (reset when started afresh).
- Own GPU images take an image's page address from Docker Hub, GitHub, Quay or NVIDIA NGC, with or without https://, and show which image they'll run.
- When an image can't be pulled, the message says whether the registry has no such image, is limiting pulls or can't be reached.
- Own GPU images take a Docker Hub page address or a docker pull command, and show which image they'll run.
- The GPU node installer restarts the machine at the end when the NVIDIA driver needs it to load.
- Servers can't be suspended or terminated while they're moving, and a cancel that comes too late finishes the move instead.
- Wrong two-factor codes are limited per day, and a lockout emails the account's owner; staff managers hear about new staff, role changes and panel API keys.
- The audit log is tamper-evident, copied to the system journal and checked with vorticpanel audit-verify.
- Nodes keep servers' disks out of their own disk scanning, and folders a GPU image declares as volumes count toward the instance.
- The Updates page warns when the update server offers an older version than before.
- Customers → SSH keys lists every customer's saved SSH keys, to search them and remove one from an account.
- Billing panels keep changing packages, addresses, backup plans and power while the license needs attention.
- Imported servers keep the virtual hardware they had, and VirtFusion customers their time zone.
- Resellers are told when any change takes a quota past 90%.
- Security fixes from a full review, and help text that matches what the panel does.
- Imported servers keep their own CPU and memory when the package they're mapped to is smaller, as they already kept their disk size.
- Debian 12 and other Linux servers with a minimal cloud kernel now read their first-boot settings, so they come up with their hostname, password, network and guest agent.
- A reinstall's snapshot is kept when a later step of the reinstall fails.
- A reinstall's snapshot is marked as not consistent when the server didn't shut down in time and was forced off.
- Imports from VirtFusion count each server's traffic the way VirtFusion did (both ways, inbound only or outbound only), and its first traffic period ends when it would have there.
- Packages created from VirtFusion's count and reset traffic the way their servers did.
- VirtFusion servers protected from deletion arrive with delete protection on.
- The import's dry run names servers that had firewall rules on the source panel, which don't come across.
- Imports can create a source panel's packages here, with the same CPU, memory, disk, traffic and port speeds, one at a time or all at once.
- A package the source panel no longer sells comes over hidden.
- Imports connect to VirtFusion panels that only answer on their own domain name.
- A VirtFusion that refuses its IP address says to use its domain name instead.
- Imports wait and try again when the old panel asks them to slow down.
- Imports say what refused the API key.
- A node can be drained while it's offline.
- Backups and snapshots of running servers are made while they keep running.
- A backup no longer fails when a node briefly loses its connection.
- Rocky Linux and AlmaLinux nodes set up their network bridge themselves.
- Moves with a restart work between AlmaLinux, Rocky, Debian and Ubuntu nodes.
- Catalog OS templates pick up newer builds by themselves.
- Windows servers answer ping over IPv6.
- IPv6 pools of any size, handing each server a block of the size you choose, down to one address.
- Packages can allow more than one storage tier.
- A banner on the server page when changes wait for a restart.
- Removing a disk waits for the server to let go of it.
- Default virtual hardware for new servers, panel-wide or per package.
- Windows sees disks on SCSI or SATA as solid-state drives.
- Nodes show physical cores and threads.
- A node keeps its place in the panel after its OS is reinstalled.
- Servers whose disks were lost can be restored from their latest backups, all at once.
- Moves get through when a node's SSH server is busy.
- Extra traffic added this month can be taken back.
- Saved Windows images get a fresh identity for each server made from them.
- Set up the bridge on a node's Settings tab.
- Staff can set a server's own monthly traffic allowance.
- Imported servers keep their own traffic limit every month.
- Pick which servers an import brings, with search and filters.
- Imported servers keep their power state, suspension, reverse DNS, traffic, extra disks and their owners' SSH keys.
- Customers get a "Your server has moved" email.
- Firmer limits on self-service servers, private networks and GPU power actions.
- Fixed moves between nodes failing with "Permission denied".
- A node's management address can be changed on its Settings tab.
- New IPv4 pools take the gateway, netmask and usable range your provider gave you.
- The fleet overview works on a new panel before any package exists.
- A node being installed isn't reported as down before it first connects.
- Consoles keep working on a node whose DNS stops answering, and the node repairs its DNS settings.
- Adopt-in-place imports show only the settings that apply to them.
- Customers get an email when their password, two-factor or keys change.
- Changing package adds the new package's IP addresses.
- Unmetered packages are never slowed or suspended for traffic.
- Staff see a line at the top of the page when an update is out.
- Hidden packages are offered only to staff and billing.
- Clearer wording across the panel.
- Moves with a restart are off only for the restart itself.
- Customers can run their own Docker image on GPU offers that allow it.
- A separate logo for the dark theme.
- Servers that can't move live move with one restart, even between Intel and AMD nodes.
- JupyterLab on GPU instances can be served over HTTPS through the panel.
- Enrolling a node is its own permission.
- Staff are alerted when a node's credential is replaced, or when a node reports more hardware than before.
- Fixes from a full security review.
- Emptying a node can give servers new addresses where theirs won't work, and email the customers.
- A server's storage type no longer stops it moving.
- Imported servers start with BIOS or UEFI to match their disk.
- Force off during a hardware change leaves the server off, and the change applies at its next start.
- Imports copy servers while they run, and each is offline only for a last short pass.
- Imports can be scheduled up to 30 days ahead, with an email to each customer.
- Notifications can be filtered by topic.
- Delete protection and hard reset for servers.
- Outbound firewall rules for customers.
- Separate inbound and outbound port speeds, and a choice of how traffic is counted.
- Webhooks to Slack, Discord and Telegram, and Prometheus metrics.
- Your own logo, and a time zone for each person.
- Any folder on a node, such as one on a second drive, can be storage.
- Floating IPs can only be reserved where the account has a server.
- IP pools can be renamed, and single addresses taken out of them.
- Staff can release a customer's floating IP.
- Staff are told when storage runs low or goes missing.
- Floating IPs come from another pool in the location when the first is full.
- Smaller fixes to support sessions, node keys and messages.
- Nodes only take orders signed by your panel.
- Private networks between nodes are encrypted with WireGuard.
- GPU instances run in user namespaces, so root inside isn't root on the host.
- GPU instances can wait for the customer to choose an image.
- Fixes from a full security review.
- Every server's backups and snapshots in one place, under Fleet.
- Report a problem sends an issue to support along with the panel's diagnostics, which you see first.
- Billing can add a customer's SSH key from their order.
- Works with the VorticPanel WHMCS module, from Downloads in your account.
- Billing can sign customers straight in to the panel.
- A log of every API call, under Administration → API requests.
- Customers and resellers can be deleted once nothing of theirs is running.
- Panel API keys belong to the panel, not a staff member, so billing keeps working when staff leave.
- The installer gets Caddy from the system's packages or Caddy's own release.
- Servers on a node that isn't answering show their status as Unknown.
- A node that stays connected but stops reporting shows as offline after 5 minutes.
- A bot check (Cloudflare Turnstile) can be turned on for sign-in and password resets.
- Resellers' own servers and the ones they sell are listed apart.
- Resellers see everything they sell in one list, with search and filters.
- Resellers can open the console and JupyterLab on their customers' GPU instances.
- A reseller's page says which of its locations can't take servers right now.
- The GPU offers list shows where each offer is sold and how much room is left.
- Resellers can sell GPU instances, limited to the offers and number of cards you give them.
- Low and high CPU priority now give half and double a normal server's share.
- The Updates page lists recent releases and this panel's update history.
- Usage reports show past months' traffic.
- Force off isn't offered while a server is moving.
- Imports bring Windows servers on the virtual hardware they had, so they boot with their own drivers.
- IDE can be chosen as a disk bus for older Windows servers.
- Imports bring UEFI servers, with or without Secure Boot, along with their boot entries.
- Encrypted copies of the controller can go to S3 or SFTP every hour or day, under Settings → Backups.
- The installer sets up a firewall on the controller: SSH, HTTP and HTTPS only.
- Pages left open across an update load again without a refresh.
- Firmer limits on requests, consoles and sign-ins.
- Staff see only what their role allows, including in support sessions.
- GPU container hosts need NVIDIA Container Toolkit 1.17.8 or later.
- Install with one command from vorticpanel.com.
- A new panel asks for its license key, with links to the free trial and the plans.
- Fixed servers on the same node sometimes failing to apply their settings at the same moment.
- The panel checks its license and looks for new versions every hour.
- Updates show each step as it happens.
- Updates no longer stop when an outside package repository is down.
- Agent updates are signed with the release key; a node that has had a signed agent only takes signed ones.
- Security fixes: two-factor guessing, root passwords no longer kept with repeated requests, import address and adoption checks, reverse DNS sync, private network guard, move keys expire, IPv6 router spoofing, GPU instance isolation.