Provider Quickstart
Get started in 5 min
Two ways to become a provider; today Path B is the one that is open. Path A, an e-mail or Google account with nothing to register and nothing to buy first, is not open yet. Want to know how much you'll earn? See Rewards.
What GPU models can I bring to this subnet?
The validator scores a fixed allow-list of NVIDIA models — Hopper / Blackwell datacenter (H100, H200, H800, B200, B300), Ampere & inference datacenter (A100, A10, T4, V100, P100, P40, M40), workstation (RTX 6000 / 5880 / 5000 Ada, RTX PRO 4000/5000/6000 Blackwell, RTX A-series, L4 / L40 / L40S, Quadro & TITAN), and consumer (full RTX 50/40/30/20-series and GTX 10/16-series). The complete list with per-model reward weight, the unrented-pool eligibility rule, and CVM-eligible GPUs lives in Architecture → Supported GPUs. The source-of-truth is GPU_MODEL_RATES in lium-io.
Two paths
| Path A: account with e-mail or Google (coming) | Path B: your own registered hotkey (open today) | |
|---|---|---|
| What you need | An e-mail address or a Google account | A Bittensor wallet, TAO for the subnet 51 registration fee, btcli |
| Steps | Create an account → add a node → it is listed | Register on subnet 51 → sign in with the hotkey → add a node |
| How you are paid | Your nodes' share of rentals is attributed to your account per UTC day once settled, in USD; unrented time is not paid on this path; withdrawals open when stablecoin payouts launch, and we e-mail you then | Rental share to your coldkey, emission on-chain — see Earnings → Payouts |
Path A — create an account, nothing to buy first
Path A is built but off in production until Lium's pool has its own registered key. Until then provider.lium.io/login shows the hotkey sign-in and provider.lium.io/signup says Account signup is not open yet. Start with Path B. The steps below describe Path A as it will open.
- Go to provider.lium.io/signup. Continue with Google, or enter an e-mail and a password (8 to 72 characters), open the confirmation link we send you and enter that password again. Either way you land in the portal signed in; it first offers to connect Discord, which you can skip.
- Install Sysbox on the GPU host (Node Quickstart step 1). Then, on Overview (the first sidebar entry of an e-mail or Google account), copy the command under Set up a GPU host and run it on the host — it is the Node Quickstart installer with the
-kvalue already filled in (Lium's listing key, the same for every such account — the node becomes yours when you add it in the portal), so you type nothing but the ports it asks for. Then click Add Node and enter the GPU type, count, address and port the installer printed, and your price per GPU (USD/hr). Lium lists your nodes for you — there is nothing to register and nothing to buy first. - Your node is checked automatically and listed for rent; from then on it earns its share of every rental. The Overview shows this month's and all-time totals.
What your host needs
- Ubuntu 24.04, or Ubuntu 22.04 with the HWE kernel; the NVIDIA driver installed (
nvidia-smiworks); Docker running; Sysbox installed. Its installer checks the host first and prints the fix for anything missing. - At least 100 Mbit/s download, a public IP address and open ports (
8080and2200by default). - At least 8 GB of RAM and 150 GB of disk.
- On an HGX H100 or H200 host: NVIDIA Fabric Manager running before you add the node (Troubleshooting).
- Current node software:
docker compose pull, thendocker compose up -d.
The full list is the Node Quickstart requirements.
How a node is checked
Every node is checked automatically when you add it and again on a regular cycle while it is listed: the GPU model and count, network speed, the Sysbox runtime and storage. A new node shows VALIDATION PENDING, then AVAILABLE once it passes, usually within about 20 minutes. A node that fails a check is not listed, and its page in the portal names the reason; Troubleshooting explains each one. Your account itself is checked once: Google must report your address as verified, or you open the confirmation link we e-mail you.
How you are paid
- Your share: 95% of what renters pay for your node; Lium keeps 5%. It is booked in USD per node and UTC day once the day is settled, and the Overview shows the running totals.
- Rentals only: time your node waits unrented is not paid on this path.
- Failed rentals: when a rental fails because of a problem on your node, the renter is refunded and your share of that billing is taken back from your balance (Penalties lists how to avoid it).
- Withdrawals open when stablecoin payouts launch, and we e-mail you then. From then on you set a USDT (US-dollar stablecoin) address in Settings → Payout, and a balance of $5 or more is paid out daily. The card shows the schedule and any fee before you save the address. A new or changed address is paid only after a 72-hour hold.
Hosts we need most
Renters most often ask for whole 8-GPU hosts: 8× H100, 8× H200, 8× B200 and 8× B300. A host of that shape fills more customer requests than the same GPUs split across smaller machines. Every model on the supported list can be listed.
The two paths do not mix: an e-mail or Google account never signs in to an existing provider's account, even with the same e-mail address. Everything from here to Next steps is Path B.
Path B — bring your own registered hotkey
Run btcli on your secure local machine, not on the provider server. Only hotkeys belong on provider infrastructure — never your coldkey.
Before you start — wallet & TAO
You need a Bittensor wallet (one coldkey + at least one hotkey) and enough TAO on the coldkey to cover the dynamic subnet 51 registration fee.
- Don't have a wallet yet? Install
btcli(pip install bittensor) and create one — see the Bittensor wallet docs for thewallet new_coldkey/new_hotkeyflow. - Need TAO? Acquire it on a supported exchange and transfer it to your coldkey's SS58 address. The registration fee is dynamic — check the live cost on the TaoMarketCap Subnet 51 registration page before submitting (or run
btcli subnet list --netuid 51).
Step 1 — Register your provider on Subnet 51
btcli subnet register --netuid 51 --wallet.name <wallet> --wallet.hotkey <hotkey>
Registration burns the dynamic subnet 51 fee from your coldkey balance. The transaction either confirms (your hotkey now has a UID on subnet 51) or fails (insufficient TAO, or the burn cost spiked between command issue and chain inclusion).
Step 2 — Sign in to the Provider Portal
- Add your provider hotkey to a supported wallet extension. Any SS58-compatible browser extension works — common choices are the Polkadot.js extension and the Bittensor Wallet extension.
- Sign in at provider.lium.io/signature-login with that hotkey.
- In Profile Settings → Central Miner Configuration, the Central Miner Server switch is ON for a new account: Lium's central miner answers the validators for your hotkey and serves the nodes you add in the portal. Leave it ON.
- In the same settings page, connect Discord. Your nodes earn subnet incentive (idle pay and the rented extra) only while your Discord account is connected; stay in the Lium Discord server with it. Rental fees are paid either way.
With the switch ON, Lium operates the CPU coordinator for you and you skip the self-hosted provider setup. Prefer to run your own? Turn the switch OFF and follow Provider Configuration for the self-hosted path.
Step 3 — Set up your first node
A provider with no GPU nodes earns nothing. Bring one online with the Node Quickstart — three steps: install Sysbox (required — validators reject any node without the sysbox-runc runtime), run the one-line mine.sh installer, and register the node back in the Provider Portal.
You're live 🎉
Being live means the provider is registered and the node is reachable. Earning starts after validators score your node — a new node shows VALIDATION PENDING first, and measured over the 175 nodes added in the 30 days to 6 Sep 2026 it turned AVAILABLE after 21 minutes at the median (64 minutes at the 90th percentile; never under 8). Meaningful rental revenue then depends on actual customer rentals matching your GPU type and price. Track scoring in the Provider Portal monitoring view and emission externally on TaoMarketCap.
Checklist: what costs you money if it is missing
The portal Overview checks these for your account and nodes under Things that cost you money if they are missing.
| If this is missing | What it costs | Fix |
|---|---|---|
| Discord connected in the Provider Portal | No subnet incentive from either pool, rented or idle | Profile Settings → Discord → Connect Discord; stay in the server (Connect Discord) |
| A provider server for your nodes | No node is checked, so none is listed or paid | Profile Settings → Central Miner Server → ON: Use Lium.io server, or keep a self-hosted provider running |
| Sysbox runtime on the host | The node is never listed | Sysbox |
| At least 100 Mbit/s download | The check fails with VERIFYX_FAILED_NETWORK_SPEED_TOO_SLOW and the node is not listed | A host with a faster uplink |
| Disk of at least 1.5 × the GPU memory (Docker data folder on that disk) | No idle pay | Move Docker's data root to the large disk before you start the node |
| Price at or under the model's soft limit | Priced above it, the node gets no idle pay; rentals still pay | Look up the model's Soft Limit on the Rewards Calculator, then lower the node's price to it or below |
| On an 8-GPU H200, B200 or B300 node: GPU splitting, GPU profiling or a confidential VM | No idle pay; rentals still pay | Any one of the three: GPU splitting (a minimum GPU count for rental below 8; the node keeps all 8 GPUs; needs XFS Docker storage), GPU profiling (ncu), or a confidential VM |
| Node software up to date | No subnet incentive, rented or idle | docker compose pull, then docker compose up -d |
| Penalties under 20% of your 14-day rental income | Every node moves to Spot: no incentive, rented or idle | Keep nodes up through rentals (Penalties) |
Rental fees arrive on your coldkey once a day, and incentive arrives on your hotkey every tempo; the schedule is on Payouts.
Next steps
- Node Quickstart — bring your first GPU online
- Connect Discord — provider Discord setup
- Provider Configuration — self-hosted vs Central Provider Server trade-offs
- Self-hosted provider setup — full install if you opted out of Central Provider Server
- Architecture — mental model of the system
- Provider Portal overview — day-2 management UI