Every tier bundles twin storage, sandbox hours, agents, awake hours and storage. Go past them and you pay the metered rate, shown right here. Sleeping agents and stopped sandboxes cost nothing but disk.
Local and Docker agents, forever free. One agent in the cloud to try it.
Your agents keep working after you close the lid.
A whole crew of agents and a twin that never forgets.
For experts and teams who publish agents to the world.
| Feature | Fan | Plus | Pro | Studio |
|---|---|---|---|---|
| Hosted agents | ||||
| Hosted agents | 1 | 3 | 10 | 30 |
| Awake hours / month | 10 | 60 | 250 | 1,000 |
| Largest machine size | Small · 1 vCPU · 1 GB RAM | Medium · 2 vCPU · 2 GB RAM | Large · 4 vCPU · 8 GB RAM | Large · 4 vCPU · 8 GB RAM |
| Storage | 1 GB | 10 GB | 50 GB | 250 GB |
| Scheduled jobs and webhooks | — | — | ✓ | ✓ |
| Digital Twin | ||||
| Digital Twin backup | — | 20 GB | 100 GB | 500 GB |
| Twin in your browser, hours / month | — | 10 | 40 | 150 |
| Checkpoints | 1 | 5 | 30 | 100 |
| Sandboxes | ||||
| Sandboxes at once | 1 | 2 | 5 | 20 |
| Sandbox hours / month | 2 | 10 | 40 | 120 |
| Largest sandbox | Small · 1 vCPU · 2 GB RAM | Medium · 2 vCPU · 4 GB RAM | Large · 4 vCPU · 8 GB RAM | Large · 4 vCPU · 8 GB RAM |
| Start from a copy of your twin | — | ✓ | ✓ | ✓ |
| Full desktop in the browser | coming | coming | coming | coming |
| Private GPU | ||||
| GPU access | — | ✓ | ✓ | ✓ |
| GPU discount | — | — | 10% off | 15% off |
| Presets and team | ||||
| Preset catalog | Community | All, incl. expert-tuned | All, incl. expert-tuned | All, incl. expert-tuned |
| Seats | 1 | 1 | 1 | 5 |
| Publish expert presets | — | — | — | ✓ |
| Web apps | ||||
| Deploy to your own domain and GitHub | coming | coming | coming | coming |
| Support | Community | Email + Discord | Priority | Dedicated onboarding |
| Price | ||||
| Monthly | $0 | $5 | $15 | $49 |
| Yearly | $0 | $50 (2 months free) | $150 (2 months free) | $490 (2 months free) |
Metered usage appears on the same invoice as your subscription. It only bills when a card is on file; the Fan tier without a card simply stops at its quota until next month.
| Unit | What it means | Price | Billed |
|---|---|---|---|
| Awake hour | one hosted agent awake for one hour, beyond your tier's quota | $0.10 | per hour |
| Agent storage | agent homes on disk, prorated by the hour | $0.05 | per GB-month |
| Twin storage | digital-twin snapshots and checkpoints, prorated by the hour | $0.05 | per GB-month |
| Twin in your browser | your twin running as a cloud computer, beyond your tier's hours | $0.16 | per hour |
| Sandbox hour, Small | 1 vCPU · 2 GB RAM · 10 GB disk, running, beyond your tier's sandbox hours | $0.09 | per hour |
| Sandbox hour, Medium | 2 vCPU · 4 GB RAM · 20 GB disk, running, beyond your tier's sandbox hours | $0.18 | per hour |
| Sandbox hour, Large | 4 vCPU · 8 GB RAM · 40 GB disk, running, beyond your tier's sandbox hours | $0.35 | per hour |
| GPU minute | private GPU machine running, boot to shutdown | see table below | per minute, minus your tier discount |
One open-weights model per machine, on a machine only you can reach. Pro takes 10% off every rate below, Studio 15%. Idle machines stop themselves. How it works.
| Machine | VRAM | Per hour | Per minute | Suggested model |
|---|---|---|---|---|
| Starter 16 GB | 16 GB | $0.75 | $0.0125 | Qwen/Qwen3-8B |
| Compact 24 GB | 24 GB | $1.00 | $0.0167 | Qwen/Qwen3-8B |
| Standard 24 GB | 24 GB | $1.40 | $0.0233 | Qwen/Qwen3-14B |
| Performance 48 GB | 48 GB | $2.45 | $0.0408 | Qwen/Qwen3-32B |
| Pro 80 GB | 80 GB | $3.15 | $0.0525 | Qwen/Qwen3-32B |
| Frontier 80 GB | 80 GB | $4.95 | $0.0825 | NousResearch/Hermes-4-70B |
| Frontier 141 GB | 141 GB | $5.70 | $0.0950 | NousResearch/Hermes-4-70B |
| Titan 180 GB | 180 GB | $7.80 | $0.1300 | Qwen/Qwen3-235B-A22B-Instruct-2507-FP8 |
Suggested models are what fits comfortably at a useful context length; you can pick any model that fits the VRAM. Per-minute price is the hourly price divided by 60 before your tier discount.
An agent that isn't chatting or running a job is frozen. Frozen machines cost us cents of storage, so they cost you nothing.
Thousands of small machines and a shared pool of GPUs, run as one cloud. You get the bulk rate without the bulk commitment.
No demos, no quotes, no enterprise tier with a "contact us" button. The price on this page is the price.
We didn't build a client; the community did, and it's free. We only sell the machines it can point at.
One hour of one hosted agent being awake: chatting with you or running a job. We meter from wake to sleep, prorated by the minute. Idle machines are put to sleep automatically.
No. Fan is free, no card. When you hit the 10 awake hours or 1 GB, the agent stops until the next month, or until you add a card or upgrade.
Once your tier's monthly quota is used, each extra unit is priced at the metered rate above and added to your next invoice. Storage is prorated by the hour, so a 12 GB day costs a fraction of a GB-month.
Yes. Every GPU machine has a maximum runtime and an idle auto-stop. Hosted agents go to sleep on idle, sandboxes and twins open in the browser stop after 30 idle minutes, and you can cap a twin's cloud hours from the plugin.
Pay ten months, get twelve. Switch from monthly to yearly any time from the billing dashboard; the remaining month is credited.
Subscriptions are refundable within 14 days if you haven't used more than the Fan quota. Metered usage is not refundable, because we already paid for it.
Your tier drops to Fan at the end of the period. Machines above the Fan quota are put to sleep, kept for 30 days, then deleted with their storage. Download or checkpoint first.