Features
Everything FlashVMM does
Instant, isolated FlashVMs for every AI task, on macOS, Linux and Windows. Pick a feature to see what it is, why it matters, how to use it and what you get.
Boot & run
Instant bootA full virtual machine with its own Linux kernel, running in 18 ms.Explore Instant boot →
NetworkingGive a VM a network card in one flag: outbound internet, checked HTTPS, every connection under your rules.Explore Networking →
Run commandsRun any command inside a VM and stream its output back, with no SSH and no network.Explore Run commands →
Files in and outCopy files and whole folders into and out of a running VM, permissions kept.Explore Files in and out →
Folder sharingShare exactly one host folder into a VM, live in both directions.Explore Folder sharing →
Guest bundlesA kernel and root filesystem in one bundle: inspect it, size its disk, run it anywhere.Explore Guest bundles →
Control & limits
Live limitsChange a running VM's CPU, network speed and disk caps with no restart.Explore Live limits →
Memory on demandTake memory back from a running VM and give it back again, live.Explore Memory on demand →
Pause and resumeFreeze a VM completely in place and carry on exactly where it stopped.Explore Pause and resume →
Run-length limitsTell a VM how long and how much CPU it may use; it stops itself when it reaches either.Explore Run-length limits →
ObservabilitySee a VM's settings, logs and live CPU, memory and network from outside, without logging in.Explore Observability →
Security
Network wallsDecide exactly what a VM may reach, see every connection it made, and change the rules live.Explore Network walls →
Hidden API keysA VM uses your API key without ever seeing it.Explore Hidden API keys →
Change trackingOne call returns a command's exit code, its output and every file it changed.Explore Change tracking →
Speed at scale
Warm poolsVMs booted ahead of time and handed to a job in 1.2 ms.Explore Warm pools →
NIC poolsNetwork cards built ahead of time, so every VM starts already online.Explore NIC pools →
GPU poolsOne model on the host GPU, shared by many VMs at once.Explore GPU pools →
State & time travel
Snapshots and clonesSave a running VM to a file and bring it back, mid-program, in 7.8 ms, or start many copies from it.Explore Snapshots and clones →
Sleep and wakePut a VM to sleep at zero memory and zero CPU, and wake it in 8.4 ms.Explore Sleep and wake →
Live forkCopy a running VM, memory, disk and processes, into new VMs in 36 ms.Explore Live fork →
DiffSee exactly which files differ between two VMs, with line-by-line changes.Explore Diff →
Checkpoints, rewind and branchEvery command becomes an undo point: go back to any step, or try several paths from it.Explore Checkpoints, rewind and branch →
Fleet
Host capacityOne view of every VM on a machine, with room checked before any VM starts.Explore Host capacity →
Fleet controllerPlace VMs across many machines by best fit, from one controller.Explore Fleet controller →
Live migrationMove a running VM to another machine and it carries on where it was.Explore Live migration →
AI & integration
MCP for AI agentsGive an AI assistant the same controls as the CLI, through 21 MCP tools.Explore MCP for AI agents →
Guest messagingCall a program inside a VM like a web service, or message it on a topic, with no network card.Explore Guest messaging →
Self-updateSigned, verified updates that install themselves without touching running VMs.Explore Self-update →
The live demoEvery feature, live, in one command.Explore The live demo →
See every feature live: flash demo