PreviewThe board below runs on sample data. Names, addresses and amounts are made up.
The network terminal. Every node, ranked by what it earned on chain.
A node's earnings, redeems, channels, bond and listing date sit in its own account on Solana, so nobody keeps this board but the chain. Requests stay between a router and a node; what lands on chain is the money.
Solana mainnet right now1 node · 22 channels · $0.017342 settled · program 7UHe…ytZS
Real accounts on mainnet, read live: the site's demo router and the seed node.
- Nodes listed
- 12
- Models served
- 9
- Paid to nodes, 24h
- $58.45
- Redeems, 24h
- 256
- Channels open
- 1,157
- Median first token
- 242 ms
- Program cut
- 0%
| # | Node | Serves | Earned, 24h | All time | Redeems | Channels | First token |
|---|---|---|---|---|---|---|---|
| 01 | llama-3.3-70b+3 | $11.46 | $233.60 | 1,557 | 234 | 182 ms | |
| 02 | llama-3.3-70b+2 | $9.53 | $185.67 | 1,565 | 213 | 236 ms | |
| 03 | deepseek-v3+2 | $9.46 | $157.08 | 925 | 187 | 171 ms | |
| 04 | qwen3-32b+1 | $6.16 | $138.85 | 1,084 | 95 | 312 ms | |
| 05 | gpt-oss-120b+1 | $5.85 | $77.20 | 545 | 115 | 204 ms | |
| 06 | gemma-3-27b+1 | $4.97 | $105.54 | 742 | 82 | 258 ms | |
| 07 | llama-3.3-70b+2 | $4.10 | $50.02 | 310 | 88 | 247 ms | |
| 08 | qwen3-coder-30b+1 | $2.72 | $22.10 | 161 | 48 | 289 ms | |
| 09 | qwen3-32b+2 | $2.07 | $10.43 | 120 | 43 | 221 ms | |
| 10 | llama-3.1-8b+1 | $1.05 | $3.09 | 59 | 23 | 274 ms | |
| 11 | llama-3.1-8b+1 | $0.7274 | $1.70 | 37 | 17 | 331 ms | |
| 12 | gpt-oss-120b+1 | $0.3437 | $0.3437 | 6 | 12 | 197 ms |
Models on the network
| Model | Nodes | Lowest, per M in / out | Cached in from | Fastest first token |
|---|---|---|---|---|
| 5 | $0.10 / $0.26 | — | Ironbridge Compute · 182 ms | |
| 4 | $0.033 / $0.13 | $0.009 | Larkspur Compute · 171 ms | |
| 3 | $0.24 / $0.99 | $0.064 | Larkspur Compute · 171 ms | |
| 5 | $0.070 / $0.17 | — | Saltmarsh 4090s · 221 ms | |
| 2 | $0.10 / $0.41 | $0.026 | Larkspur Compute · 171 ms | |
| 3 | $0.19 / $0.75 | — | Moraine Inference · 247 ms | |
| 4 | $0.018 / $0.045 | — | Kestrel GPU Co-op · 236 ms | |
| 2 | $0.022 / — | — | Ironbridge Compute · 182 ms | |
| 2 | $0.011 / — | — | Saltmarsh 4090s · 221 ms |
Your row goes here. List a GPU, a Mac or a rack in one command.
It finds the models on your vLLM, SGLang or Ollama server, drafts a price list with cached-input prices, and registers the node on Solana. Routers pay you per answer, and you redeem whenever you like with nothing taken out.