[Network]

Run a node

Plug in H100s or a single 4090, stake TOPO and earn up to 85% of every token your GPU serves, paid daily in USDT.

Requirements

Component	Minimum	Recommended
GPU	1× RTX 4090, 24 GB	8× H100 or B200, 80 GB+
CPU and memory	8 cores, 64 GB	32 cores, 512 GB
Storage	1 TB NVMe	4 TB NVMe
Network	1 Gbps, public IP	10 Gbps, under 20 ms to a Topo region
Stake	2,500 TOPO per GPU	10,000 TOPO per GPU
Software	Ubuntu 22.04, NVIDIA driver 550+	Ubuntu 24.04, driver 570+

Set up

01
Pull the node image
The node ships as one container and needs the NVIDIA Container Toolkit.
bash
```
docker pull ghcr.io/topolabs/topo-node:1.8.2
```

Register your wallet

bash

docker run --rm -it ghcr.io/topolabs/topo-node:1.8.2 \  register --wallet x8f3a…c21e --region eu-west-2

03
Stake
Stake from the operator dashboard or the CLI. Stake stays locked for 14 days after you stop serving.
bash
```
topo-node stake --amount 2500 --currency topo
```

Serve

Start the node. It runs a 10 minute benchmark, and if it passes, the router starts sending traffic within the hour.

bash

docker run -d --gpus all --name topo-node \  -v /var/lib/topo:/data -p 7420:7420 \  ghcr.io/topolabs/topo-node:1.8.2 serve --config /data/node.toml

Configuration

node.toml

toml

[operator]wallet = "0x8f3a…c21e"region = "eu-west-2" [models]serve = ["deepseek-v4-pro", "qwen3.8-max", "muse-spark-1.3"]max_concurrency = 32 [payouts]currency = "usdt"chain = "base"

Earnings

Operators keep up to 85% of the bill for every token they serve, paid daily in USDT on Base or Solana. Faster, more reliable nodes win a larger share of the router's traffic.

GPU	Typical models	Est. per month
RTX 4090	Muse Spark 1.3 and small open models	$180 to $320
H100 80 GB	DeepSeek V4 Pro, Qwen3.8-Max	$1,400 to $2,100
B200	Frontier and large open models	$2,900 to $4,200

Estimates at 70% utilization and current list prices. Not a guarantee of earnings.