Application

Pistachio

A downloadable AI inference client for the Morpheus decentralized network. Run it yourself on your machine or your own server, no vendor lock-in, direct peer-to-peer connections to compute providers. There is no DRM3-hosted demo; you self-host.

Live
Access
Public
State
Live
Family
App
Tier
Application
Who it's for
Developers and operators who need censorship-resistant AI inference without vendor lock-in.
Get started
Install the binary, stake MOR, and run inference on localhost.
Releases on GitHub · no hosted demoupdated 13:24 UTC
v0.32.0
Latest release
180
Downloads
35
Releases

Morpheus is a decentralized peer-to-peer AI network where independent compute providers run models and register on-chain. The default gateway routes through a third-party broker who adds latency, takes a cut, and becomes a single point of failure. Pistachio removes that broker entirely. Your machine connects directly to the provider. One less hop, no middleman, no markup.

MOR staking is collateral, not payment. It returns in full when sessions close. The only real cost is a small amount of ETH gas on Base. Providers compete on price and uptime. You pick the model, Pistachio finds the best available provider.

Permissionless inference matters. No vendor approval, no API waitlists, no terms that change overnight. For open data processing at scale, that independence is the point. DRM3 runs millions of articles per year through AI analysis on this infrastructure. The same network is available to anyone running Pistachio.

OpenAI and Anthropic-compatible APIs on localhost. Drop it into any existing workflow. Auto-staking, auto-renewal, zero downtime. Available as a Docker image to run your own gateway on any container host. Total cost: roughly $5 to $8 per month for the container, plus whatever MOR you stake for capacity.

Screenshots

The Pistachio dashboard: providers, sessions and the local inference gateway
Pistachio on your machine: the providers it reaches, the sessions it runs, the gateway on localhost

Capabilities

Permissionless: no vendor approval, no API waitlists, no terms that change overnight

Peer-to-peer: your machine connects directly to compute providers

Zero capex: providers supply the compute, MOR collateral returns when sessions close

No broker markup: one less hop, lower latency, better economics

OpenAI + Anthropic API compatibility on localhost

Docker image to run your own gateway ($5-8/mo self-hosted)

MOR staking with automatic session renewal

NFT-gated access tiers (Free through Diamond): contracts deployed, program paused

Multi-provider routing with health monitoring

Ed25519 provenance signing on every response