ROBINHOOD CHAIN · USDG

Decentralized GPU compute
verified on-chain

Permute is a marketplace for inference and GPU capacity: providers earn USDG, users pay in USDG or $permute, and usage is visible on Robinhood Chain. Hybrid routing keeps the network live while supply grows.

How it works →Start building →

Hub preview · BetaPreview seed data — replaces with live orchestrator metrics at beta launch.Open live dashboard →

Workers online

Beta seed supply

$128

USDG / 24h

Hub preview

Robinhood Chain

Settlement

Chain 4663

34%

Network-routed

Rest API fallback

0.1

Rent GPU capacity by the hour

Browse live GPU providers in the Hub, pay in USDG on Robinhood Chain, and lock rentals in escrow — with IDs and receipts you can come back to.

0.2

Hybrid inference that stays online

Chat and API jobs route to open GPUs when they are online, labeled fallbacks keep responses flowing while the network grows.

0.3

On-chain utility, not a black box

Prepay $PERMUTE for paid models, stake for rewards, post provider bonds in USDG, and trace usage through contracts on Robinhood Chain.

For builders

Plug in once. Run inference on the mesh — not only a single vendor.

Permute gives your app a path to community GPUs on Robinhood Chain, USDG-native settlement, and capacity you can rent by the hour when you need dedicated hardware. Ship the feature; we handle routing and provider economics behind the scenes.

examples/inference-client.ts

const res = await fetch("https://permute.app/api/chat/completions", {

method: "POST",

headers: { Authorization: "Bearer permute_sk_live_…", "Content-Type": "application/json" },

body: JSON.stringify({ model: "glm-flash", messages: [{ role: "user", content: "Hello Permute" }] }),

});

const data = await res.json();

// routing: network:<gpuId> or fallback:<provider>

console.log(data.net?.routing?.source, data.choices?.[0]?.message?.content);

200 OK · application/json

{
  "object": "chat.completion",
  "model": "glm-flash",
  "choices": [
    "message": {
      "role": "assistant",
      "content": "…"
    }
  }],
  "usage": {
    "total_tokens": 842
  },
  "net": {
    "routing": {
      "source": "network:worker-a1"
    },
    "job_id": "7c4e9a2f-…",
    "settlement_tx": null
  }
}

// User Research

Designed to Solve User
Challenges

Every aspect of the platform was carefully designed to address the real challenges users face when accessing and managing computing resources.

78%

Users stated that infrastructure costs significantly impact project growth.

72%

Developers prefer platforms that provide instant access to GPU resources.

65%

Complex cloud management reduces developer productivity.

// Problem & Solution

GPU access without
opaque cloud lock-in

Hyperscaler GPUs are expensive and centralized. We connect independent providers and developers through open inference, USDG settlement on Robinhood Chain, and on-chain receipts you can audit.

Problem

AI teams need GPU inference now, but centralized APIs hide pricing, routing, and uptime — and on-chain products rarely expose real compute economics.

Solution

Permute combines a provider marketplace, hybrid inference routing, USDG payments, and $permute staking — with a public Hub dashboard for jobs, revenue, and network vs fallback transparency.

Marketplace execution
on Robinhood Chain

Providers run worker agents on their GPUs. Users consume via chat or API. Jobs settle in USDG; stake $permute for rewards and provider collateral. TEE attestation and zk proofs are on the roadmap — not required for beta.

// 01

Route inference

Prefer community GPUs when online; fall back to verified API providers so users always get a response — with source labeled in the Hub.

GPU

USDG

4663

// 02

Pay providers

Network jobs pay providers in Robinhood Chain USDG. Protocol fee and $permute staking align demand with supply as the marketplace grows toward Akash-style leases.

// 03

Verify usage

Job receipts anchor model, token count, worker, and settlement on-chain. The Network page shows revenue, routing mix, and provider heartbeat — the proof layer for TGE.

DEVELOPERS

Build on the Hub — inference today, marketplace leases next

Same stack whether you chat, call the API, or run a worker: USDG on Robinhood Chain, transparent routing, and staking utilities wired in the Hub.

Open developer console →

OpenAI-compatible API

Point existing clients at the Hub. Streaming, usage metadata, and routing source on every response.

Robinhood Chain native

USDG deposits, $permute credits, and explorer-linked receipts on chainId 4663.

inference.ts

// POST /v1/chat/completions
const res = await permute.chat({
model: 'qwen-lite',
stream: true,
payWith: 'USDG',
});

// res.meta.routing.source → 'network' | 'fallback:gemini'

Decentralized GPU computeverified on-chain