Pistachio
DRM3's Morpheus-compatible P2P inference client.
What is Pistachio?
Pistachio is a lightweight gateway that runs on your machine and connects you to AI models on the Morpheus network. You stake MOR tokens to unlock access — no API keys, no cloud accounts, no middlemen.
It exposes an OpenAI-compatible API on localhost, so any tool that speaks the OpenAI protocol works out of the box. Point Claude Code, Continue, or any other AI dev tool at it and go.
- Runs locally — your machine, your data, your keys
- OpenAI-compatible API at
localhost:19377/v1 - Built-in dashboard with chat, model browser, and session management
- CLI + headless daemon mode — run on servers, VPS, or anywhere without a GUI
- Works with Claude Code, Continue, and other AI dev tools
- Auto-manages sessions — stake once, the autostaker handles restakes
Pistachio is available as a command-line client, a headless daemon for servers, or with a full dashboard UI. Run it on your laptop, a cloud VPS, or anywhere you want permissionless AI access.
Requirements
- MOR tokens on Base
- A small amount of ETH on Base for gas
- A wallet with your private key
Install
Quick Start
Open http://localhost:19377 to access the dashboard.
Live Demo
Don't want to install yet? Try the hosted version — full dashboard, live on the Morpheus network:
⚡ Try Pistachio Live — pistachio.inference.drm3.network →Use with Claude Code
Pistachio speaks the Anthropic Messages API. Point Claude Code at your local Pistachio instance:
That's it — Claude Code will route through Morpheus providers via Pistachio. Permissionless, private, no API keys.
Use with Other AI Tools
Pistachio also exposes an OpenAI-compatible API:
Direct Download
All Releases & Downloads →How It Works
Pistachio stakes MOR tokens to open sessions with AI providers on the Morpheus network. Each session gives you one concurrent request lane to a specific model. The process is fully automated:
- You stake MOR — the SDK locks MOR in the Morpheus SessionRouter contract on Base
- A session opens — you get direct P2P access to a provider running the model
- The session expires — Pistachio automatically closes it, reclaims your MOR, and restakes
- You get your MOR back — you only pay for compute time used, unused MOR is returned
This restake cycle happens automatically in the background. You don't need to manage sessions manually.
Capacity & NFT Passes
Current release: every wallet gets 20 MOR automatically. Claim your free Starter pass for an additional +20 MOR — that's 40 MOR free to get started.
For production use — multiple models, longer sessions, less gas — you need more capacity. Pistachio Passes are NFTs on Base that increase your autostake limit:
Capacity is additive — buy multiple passes to stack. Want 40K capacity? Two Diamonds. Passes are tradeable ERC-721 NFTs on Base with a 3% royalty on secondary sales.
Get a Pistachio Pass →Why Higher Capacity Saves Money
Every restake is 3 on-chain transactions (close + withdraw + open). On Base L2, each costs about $0.004 in ETH gas. At 20 MOR, your sessions last ~23 minutes — that's a lot of restaking.
Higher capacity means bigger stakes per session, which means longer sessions, which means fewer restakes, which means less ETH burned on gas. A Silver pass running 5 models saves over $2,000/year in gas compared to the free tier.
Become a Provider
Run AI models and earn MOR by serving inference requests to Pistachio users. Providers stake MOR, register their models on-chain, and receive direct P2P connections from clients.
Learn about operating a Morpheus compute node →