Pricing

Three tiers. The first is free.

The cf CLI and the version-control verbs are free on the servers you connect (up to 50). ComputeForge cloud servers are the paid tier when the primitives need our hardware. AI features and premium services are not on sale yet — described below.

Tier 1 Free

Free. The cf CLI and the verbs that make it useful.

Server history, tags and restores, scans, Forge DNS basics, Forge Mesh basics, activity log + audit trail, the web panel and the REST API — for up to 50 connected servers with 90 days of history, at any provider, including ours. No card. The Terraform provider joins them when it ships. We want this in your shell the way git is. The verbs are free; what a verb creates on our cloud (a checkpoint, a branch server) is billed like any snapshot or server, by the second.

The cf verbs — free on the servers you connect (cf branch: our cloud)

cf log
cf diff
cf blame
cf branch
cf tag
cf checkout
cf cherry-pick
cf revert
cf show
cf pushComing soon
cf scan

Freethe cf vocabulary

Works on the Linux servers you connect — yours, ours, anyone's.

The cf CLI incl.
Server history, tags, restores, scans incl.
Forge DNS basicsincl.
Forge Mesh basics incl.
Lifeguard monitoring: up to 50 health checks per organization, every 10 seconds or slower incl.
Activity log + auditincl.
Panel + APIincl.
Terraform provider Coming soonincl.
Price$0/mo
Sign up
no card · works on servers at any provider
Tier 2 ComputeForge cloud servers — the hosting tier

ComputeForge cloud servers — the hosting tier.

Linux servers in one region, on AMD EPYC hosts with NVMe storage, billed monthly or by the second. Every server plan and add-on we sell is listed below at the price the panel charges. Object storage is $0.02 per GB a month; outbound transfer up to twice what you store each month is included, then $0.01 per GB.

Pricing The cf CLI is free. Hosting tier priced below.

Pick a plan. Pay monthly or by the second.

Pricing here is for ComputeForge cloud servers. Every plan is NVMe storage on AMD EPYC servers, with one IPv4 address and free IPv6. Pay the monthly rate, or choose per-second billing for a server: the same plan costs about 20 % more per second than the monthly rate, and billing stops the second you delete the server. Prices are in US dollars, before tax.

Cloud 1 GB

Side projects & static sites
$5/mo
  • 1 GB memory
  • 1 vCPU AMD EPYC
  • 25 GB NVMe
  • 1 TB bandwidth / month
  • 1 IPv4 + free IPv6
Choose Cloud 1 GB

Cloud 64 GB

Heavy workloads
$227/mo
  • 64 GB memory
  • 20 vCPU AMD EPYC
  • 1.6 TB NVMe
  • 7 TB bandwidth / month
  • 1 IPv4 + free IPv6
Choose Cloud 64 GB

Cloud 128 GB

Database & enterprise
$477/mo
  • 128 GB memory
  • 32 vCPU AMD EPYC
  • 3.2 TB NVMe
  • 9 TB bandwidth / month
  • 1 IPv4 + free IPv6
Choose Cloud 128 GB

Every plan

PlanvCPUMemoryNVMe diskBandwidth / monthMonthly
Cloud 1 GB11 GB25 GB1 TB$5
Cloud 2 GB12 GB65 GB2 TB$8.50
Cloud 4 GB24 GB100 GB3 TB$17
Cloud 8 GB48 GB200 GB4 TB$34
Cloud 16 GB816 GB400 GB5 TB$68
Cloud 32 GB1232 GB800 GB6 TB$117
Cloud 64 GB2064 GB1.6 TB7 TB$227
Cloud 96 GB2496 GB2.4 TB8 TB$357
Cloud 128 GB32128 GB3.2 TB9 TB$477
Cloud 192 GB48192 GB4.8 TB10 TB$727
Cloud 256 GB64256 GB6.4 TB10 TB$977

Add-onsbilled per month

Snapshots are billed on the size of the disk they copy. Backups are billed on the size of the disks they protect, whatever the retention.

Extra IPv4 address$2 / mo
IPv6 rangefree
Volumes (block storage)$0.07 / GB
Snapshots$0.05 / GB
Backups$0.05 / GB

Per-second and custom plans

Per-second billing applies to a server and its volumes, snapshots and extra IPs. A stopped server is billed like a running one; delete it to stop paying. Each line is rounded up to the cent.

Per-second ratemonthly rate + 20 %
Extra vCPU (custom plan)$3 / mo
Extra memory (custom plan)$3 / GB
Extra disk (custom plan)$0.07 / GB
Sign up
custom plan = any Cloud plan plus extras · ask us with the contact form
Tier 3 AI features + Premium services

AI features and premium services. Not on sale yet.

AI features Coming soon

AI sysadmin (anomaly explanation, fix suggestions, topology authoring), AI-driven autonomous responses, semantic config-diff understanding. LLM inference has a real marginal cost, so this tier will be metered: you will pay for the inference you use. It will work on any server you connect — your VMs at AWS, your boxes at Hetzner, our metal — wherever the agent runs.

Pricing modelpay-per-use
Where it runsany connected server

Premium services

Advanced RBAC + SSO. Higher Forge DNS / Mesh quotas for fleets at scale. Forge Orchestrator (autonomous response with declared topology). Premium support. Priced per seat or fleet. None of these is on sale yet: tell us what you need.

Pricing modelper seat / fleet
Audienceteams + enterprise
Talk to us

Per-second billing — what it actually means

"Hourly billing" at hosts like Hetzner, Linode and Vultr means: you pay for an hour even if you used the VM for two minutes. We don't do that. Choose per-second billing when you create a server, spin up a CI runner and delete it 17 seconds later, and you pay for 17 seconds. The same is true for its volumes, snapshots and extra IP addresses: you stop paying the second you delete them. Each line on the invoice is rounded up to the cent, so any use costs at least $0.01. A stopped server is billed like a running one. Bandwidth is not metered per second: each plan comes with a monthly allowance.

Per-second prices are about 20 % above the monthly rate, so for a server that runs all month the monthly plan is cheaper. For workloads you delete between jobs, per-second billing can cost much less. If your CI servers exist 200 minutes a day and you delete them between runs, you pay for 200 minutes, not 24 hours.

What's included at the headline rate

Included with every server: the plan's vCPU, memory, NVMe disk and monthly bandwidth allowance, one public IPv4 address, free IPv6, and placement on AMD EPYC hosts.

Add-ons, billed separately: volumes ($0.07 per GB a month), snapshots ($0.05 per GB a month, on the size of the disk they copy), backups ($0.05 per GB a month of the disks they protect) and extra public IPv4 addresses ($2 a month each). On a per-second server they are billed by the second at about 20 % more.

Apps are free. A server deployed from an app (Docker, a LEMP stack, Nextcloud, WordPress) costs exactly what the same server costs without it.

FAQ

Is the cf CLI really free?

Yes. The CLI, the verbs, the panel, the REST API, Forge DNS basics, Forge Mesh basics and Lifeguard monitoring are free on any server you connect. No card. No per-command charges on the verbs. The Terraform provider will be free too when it ships. We want cf to be as common as git, so we don't put the vocabulary behind a paywall. What a verb creates on our cloud — a checkpoint, a branch server — is billed like any snapshot or server, by the second.

The CLI installs with one command from ComputeForge's own servers: see Install the CLI. The agent installs from the panel (Add › Server from another provider).

What does per-second billing actually mean for my bill?

On a monthly plan you pay the monthly rate. On per-second billing a server that never stops costs about 20 % more than the monthly plan; you save when you delete servers, volumes or snapshots you no longer need, because billing stops the second they are deleted.

Can prices change?

Yes. Prices can change, including for services you already run. A new price applies from the date it takes effect, not to usage already billed. (See /terms, sections 6 and 15.)

What happens if I exceed my plan?

Bandwidth is a monthly allowance per plan, and there is no automatic overage charge. If a server keeps using far more than its allowance, we may ask you to move to a bigger plan or to pay for the excess from then on (see /terms). More disk is a volume at $0.07 per GB a month. The panel shows what you have spent this month and the projected total.

Can I bring my own VMs from another provider?

Yes — and the verbs work on them at no charge, for up to 50 connected servers. Install the agent (one command, no reboot, signed and verifiable) and the VM joins your panel. You get history, tags and file restores (on servers that allow remote changes), scans and mesh membership, plus DNS failover when a probe goes red; the migration wizard Coming soon follows. Bring fifty VMs from five providers, run them all from one panel, never spin up a ComputeForge VM if you don't want to.

How are AI features priced? Coming soon

Metered, by inference cost. LLM calls have a real marginal cost we pass through (with margin), rather than rolling them into a flat fee that subsidizes heavy users at the expense of light ones. You'll see per-feature usage in the panel and can set spend caps.