Starter
$25/mo
Annual $250/yr · 2 months free
Solo builders running Pi or Hermes 8B with a local proxy. Just enough to keep your agent's home alive 24/7.
- 1 vCPU
- 2 GiB RAM
- 10 GiB SSD
- 100 Mbit/s burst
Pricing · last updated 2026
Managed agent hosting,
priced like infrastructure should be.
One agent, one Firecracker microVM, one flat monthly bill. No metered tokens, no markup on your model spend, no surprise line items. Bring your own OpenRouter / OpenAI / Anthropic key and we get out of the way.
How much does jurniti cost?
jurniti charges a flat monthly fee per agent VM. Starter is $25/month (1 vCPU, 2 GiB RAM), Pro is $49/month (2 vCPU, 6 GiB RAM), and Max is $99/month (4 vCPU, 12 GiB RAM). Go annual and get 2 months free — $250 / $490 / $990 per year. Every plan ships with a 30-day money-back guarantee, BYOK for every model provider, and a real Firecracker microVM — not a container.
Plans
One VM per agent. Pick your size.
Flat monthly or annual prepay. Every tier is the full platform — only the box size changes.
Pro
Most popular$49/mo
Annual $490/yr · 2 months free
The default. Headroom for Hermes-3-8B at Q4, OpenClaw with a heavy plugin set, or a Pi instance you actually live in.
- 2 vCPU
- 6 GiB RAM
- 50 GiB SSD
- 200 Mbit/s burst
- Custom subdomain
Max
$99/mo
Annual $990/yr · 2 months free
For agents you'd cry about losing. Daily snapshots, room for a 12 GiB working set, and the headroom to keep multiple skills warm.
- 4 vCPU
- 12 GiB RAM
- 100 GiB SSD
- 500 Mbit/s burst
- Daily snapshot
Side-by-side
Compare the three plans.
Same platform, three box sizes.
| Starter | Pro | Max | |
|---|---|---|---|
| Monthly price | $25 | $49 | $99 |
| Annual (2 months free) | $250/yr | $490/yr | $990/yr |
| vCPU | 1 | 2 | 4 |
| RAM | 2 GiB | 6 GiB | 12 GiB |
| SSD | 10 GiB | 50 GiB | 100 GiB |
| Bandwidth | 100 Mbit/s | 200 Mbit/s | 500 Mbit/s |
| Custom subdomain | Auto-assigned | Custom | Custom |
| Daily snapshot | — | — | Included |
| Firecracker microVM | Yes | Yes | Yes |
| BYOK for every model | Yes | Yes | Yes |
| 30-day money-back | Yes | Yes | Yes |
On every plan
What's included before you spend a cent.
The platform doesn't change between tiers. Only the box does.
Real Firecracker microVM
Every agent gets its own KVM-isolated virtual machine. Hardware boundaries, not namespace tricks. The same primitive AWS uses for Lambda.
BYOK for every model provider
OpenRouter, OpenAI, Anthropic, Nous Portal — paste your key into the VM once and your tokens never leave it. We don't proxy, don't see counts, don't mark up.
Public TLS subdomain
<you>.jurniti.com is auto-provisioned and certificate-managed. Plug it into your agent's webhook, dashboard, or chat bridge.
Persistent /persist volume
Skills, memory, conversation history, plugin state — anything you write under /persist survives restart, resize, and snapshot.
30-day money-back guarantee
Cancel your first plan within 30 days for a full refund, no forms, no support gauntlet. The guarantee covers your first subscription and ends once you expand. After that, cancel anytime — we pro-rate the rest of the month back.
Template + fork
Snapshot any VM into a template (private or public). Fork a public template to spin up a configured agent in one Checkout.
What you won't see on the bill
The costs we deliberately don't charge.
Most managed-agent billing pages are minefields. This one isn't.
Token markup
Your model bill goes to your model provider, not to us. BYOK is non-negotiable; we never insert ourselves into the largest line item on your bill.
Per-request fees
No usage metering. No per-invocation charge. Run your agent into the ground at no incremental cost.
Idle-time charges
The VM is yours 24/7. We don't bill differently when it's busy and we don't auto-suspend it for being quiet.
Egress overages
Bandwidth is generous (100/200/500 Mbit/s burst). We don't meter outgoing bytes.
Snapshot fees
One free template per active subscription. Additional private templates are $1/mo each. Public templates with traction are always free.
Onboarding / setup
Auto-provision runs the moment Stripe confirms the payment. No founder-led demo, no setup tax, no onboarding call required to get root.
FAQ
Questions you should ask before paying anyone.
- How much does managed agent hosting cost on jurniti?
- Starter is $25/month or $250/year (2 months free). Pro is $49/month or $490/year. Max is $99/month or $990/year. Every plan is a flat fee per VM with no token markup, no metered surprises, and a 30-day money-back guarantee.
- Is there a free trial or free tier?
- No. Spinning up a real Firecracker microVM costs us roughly $0.30–$1.20 a day depending on plan, so a free trial would mean burning cash on browsers who never convert. Instead, every plan ships with a 30-day money-back guarantee on your first plan: cancel within 30 days for a full refund, no questions asked. The guarantee applies to your first subscription and ends once you expand (add a second VM, upgrade, or buy a Templates plan) — by then you've decided we're worth it. You can always cancel anytime; after 30 days we pro-rate the rest of the month back to you.
- Why a microVM instead of a Docker container?
- Docker shares one Linux kernel across every tenant on the box. A kernel bug or container escape in your neighbor's workload can become your problem. Firecracker spins up a real KVM-backed virtual machine per tenant — the same primitive AWS uses to isolate Lambda invocations. Hardware-enforced boundaries, not namespace tricks.
- What does BYOK mean in practice?
- Bring Your Own Key. Your OpenRouter, OpenAI, Anthropic, or Nous Portal key is pasted once into your VM's environment file and never leaves it. We don't proxy your model traffic, don't see your token counts, and don't mark up your spend. You pay your model provider direct — typically the largest line item on your bill, and one we have no business inserting ourselves into.
- Can I run multiple agents on one VM?
- No — each agent gets its own microVM and its own subscription. Hermes + OpenClaw + Pi at Starter is 3 × $25/month. We don't bundle harnesses inside one VM because it would break the per-tenant hardware isolation boundary and starve resources at the Starter tier. Each VM is billed independently, so you can mix monthly and annual across agents.
- Where do my VMs run?
- us-west-2 today, on AWS EC2 nested-virtualization hosts — one density-sliced host per region. Additional regions land when committed regional MRR clears the cost of a second box; we rent capacity ahead of demand only when the math works, so we never have to raise prices because of idle infrastructure.
- What happens to my data if I cancel?
- Your persist volume (the /dev/vdb mount that holds your skills, memory, and agent state) stays live until the end of your paid period. Before deletion you can fork it into a private template — that creates a snapshot you can restore into a fresh VM later or share publicly. After deletion we keep no backups; the boundary is real.
- Can I switch plans?
- Yes. Upgrade or downgrade from the dashboard; Stripe pro-rates the difference. Resizing a VM happens at the next restart (a few seconds of downtime). Your persist volume comes with you across the resize.
- Do you charge for templates or forks?
- Public templates are unlimited and free forever — that's the growth loop. Private templates use a count-based Templates plan: 4 tiers (t5/t10/t20/t50 = 5/10/20/50 private slots) at $5/$9/$17/$29 per month, billed as one SubscriptionItem on your oldest active sub. There is no free private tier. Forking a public template into a new VM is a normal Stripe Checkout — same plan price, no fork fee.
Pick a harness, pick a plan,
be running in three minutes.
Auto-provision runs the moment Stripe confirms your payment. Your VM, your subdomain, and your shell are waiting when the welcome email lands.
Or browse community templates to fork a configured agent