Route peer-to-peer AI models through your laptop's AntStation directly into your VPS-hosted Hermes agent — no AntStation installation on the server needed.
AntStation connects you to a decentralized network of AI providers — peers offering models at competitive P2P pricing. Running it on your laptop and tunneling to your VPS gives Hermes access to that entire network without any server-side installation.
The short version above assumes you already have USDC deposited. For a full walkthrough of depositing funds via Rabby, MetaMask, or Coinbase Wallet using AntStation's local web UI (http://localhost:3118), see the Complete Mac Guide.
The AntStation desktop app doesn't expose CLI commands by default. Install the CLI:
npm install -g @antseed/cli
Verify it works:
antseed --version
AntStation won't route requests until you pin a peer — think of it like selecting a VPN server.
Browse available peers:
antseed network browse
This shows all available providers with their models, pricing, and reputation.
Inspect a specific peer for details:
antseed network peer <peer-id>
For example, to see Surplus Intelligence's full service list (my preferred peer — extremely competitive pricing for the model quality, plus 75+ free models):
antseed network peer 0e49122e76bd8b9ccb2fe10c0088c41ceb608927
Pin your chosen peer:
antseed buyer connection set --peer 0e49122e76bd8b9ccb2fe10c0088c41ceb608927
Verify the pin:
antseed buyer status
You should see:
Pinned peer │ 0e49122e76bd8b9ccb2fe10c0088c41ceb608927
Tell Hermes to use AntStation as a model provider. Give Hermes this prompt:
I want to add AntStation (a peer-to-peer AI network buyer proxy) as a model provider option when I use the /model command.
The setup details:
- AntStation buyer proxy runs on localhost:8377
- It supports OpenAI Chat Completions API at http://localhost:8377/v1
- For OpenAI-compatible requests, use:
OPENAI_BASE_URL=http://localhost:8377/v1
OPENAI_API_KEY=unused (any placeholder works)
Please:
1. Add "AntStation" as a selectable option in the /model command list
2. When selected, configure all API calls to route through http://localhost:8377
3. Use the model name from the user's request to route to the appropriate provider on the P2P network
4. Keep existing providers working alongside this new option
5. Save the configuration so it persists across sessions
Or manually add this to your ~/.hermes/config.yaml:
custom_providers:
- name: antstation
base_url: http://localhost:8377/v1
api_key: unused
api_mode: chat_completions
Add port 8377 to your existing SSH command:
ssh -L 18789:localhost:18789 -L 9119:localhost:9119 -R 11434:localhost:11434 -R 8377:localhost:8377 account@your-vps-ip
What each port does:
| Port | Direction | Purpose |
|---|---|---|
-L 18789 |
Local → VPS | Hermes/OpenClaw dashboard |
-L 9119 |
Local → VPS | OpenClaw dashboard |
-R 11434 |
VPS → Local | Hermes/OpenClaw → Ollama |
-R 8377 |
VPS → Local | Hermes/OpenClaw → AntStation (NEW) |
Keep this SSH session running in a terminal, or set up autossh / a systemd service to maintain the tunnel automatically. The tunnel is the only thing connecting your VPS to your local AntStation.
First, see what models your pinned peer offers:
antseed network peer 0e49122e76bd8b9ccb2fe10c0088c41ceb608927
Then switch to a model in Hermes (model and provider must be specified together):
/model opus-4.7 --provider antstation
Pricing varies by peer. Below is a sample model lineup at typical AntStation rates — Surplus Intelligence (my pinned peer) often undercuts these and includes 75+ free models on top:
| Model | Input $/1M | Output $/1M | Best For |
|---|---|---|---|
| gpt-5.4 | $0.25 | $1.50 | Chat, code, multimodal |
| gpt-5.5 | $0.40 | $2.00 | Chat, reasoning, multimodal |
| minimax-m2.5 | $0.21 | $0.84 | Chat, writing, creative |
| minimax-m2.7 | $0.21 | $0.84 | Chat, writing, creative |
| minimax-m2.7-highspeed | $0.42 | $1.68 | Fast chat, writing, creative |
Want to try a different provider? Browse the network, find a peer you like, and pin it:
# See all available peers
antseed network browse
# Pin a different peer (e.g., Open Forge)
antseed buyer connection set --peer 1d90f467689d499dc435e5744b4613c3203eb0aa
# Or back to Surplus Intelligence (recommended — best price/quality)
antseed buyer connection set --peer 0e49122e76bd8b9ccb2fe10c0088c41ceb608927
# Clear the pin (no default peer)
antseed buyer connection clear
Once pinned, any model you request will be routed to that peer automatically — no changes needed in Hermes.
New users get $1 trial credit. To deposit more:
antseed payments
This opens a local web interface where you can connect your wallet and add USDC.
Run:
antseed buyer connection set --peer <peer-id>
-R 8377:localhost:8377curl http://localhost:8377/v1/models from your VPSThis warning is cosmetic — just try sending a message anyway. It happens because Hermes tries to validate from the VPS before routing through the tunnel. The actual model call works fine once you send a message.
npm install -g @antseed/cliantseed network browseantseed buyer connection set --peer <peer-id>-R 8377:localhost:8377 to your SSH command/model <model-name> --provider antstationThat's it — your VPS-hosted agent now has access to the entire AntSeed P2P network, powered by whatever providers you pin.
Happy hacking!
See also: Want to understand the full AntSeed protocol — architecture, payments, reputation system, and why it matters? Read AntSeed: The Open Market for AI Inference That Has No Gatekeepers. For the Mac desktop walkthrough, check out the Complete Mac Guide.
💬 Comments