Every ParalleliX AI response runs on a live GPU node and carries a cryptographic Proof-of-Execution. No black boxes. No data centers you can't see. No kill switch a government can flip.
// demand: ParalleliX AI
ParalleliX AI creates the demand. Operators stake, run nodes, and earn streaming rewards for uptime.
Three ways in. Operators stake $PRLX and run a node for uptime rewards. ParalleliX AI users spend credits on inference. Stakers earn $ETH & $PRLX from the reward pool.
One command sets up your GPU or CPU: the Node CLI detects your hardware, prints a prefilled staking link, and configures itself when your stake lands on-chain. Rewards stream per-second for uptime, claimable any time. No slashing: your stake is always returned.
Open-source models running on network compute. Deposit $PRLX into an on-chain credits vault, spend per request, withdraw what you do not use. Every reply carries a verified Proof-of-Execution.
Stake $PRLX in the reward pool to earn streaming $ETH and $PRLX rewards. The token is renounced, owner is 0x0000. No admin keys. No inflation.
A terminal coding agent in the Claude Code style. Inference dispatched to ParalleliX GPU nodes, SHA-256 Proof-of-Execution on every call, paid in $PRLX credits. One-line install, one key, ready in a minute.
Requests in. Scheduled across the node pool. Throughput scales with the network.
ParalleliX AI streams inference requests to the coordinator. Each request carries a model name, token count, and hardware class. The coordinator queues them for scheduling.
The scheduler routes each request to a capable, online node: capability match, capacity-aware queuing, best-fit allocation. Multiple requests fan out across the pool in parallel.
Each node runs the inference request and returns the result with a SHA-256 Proof-of-Execution hash commitment over (request_id, result, node_id). Add nodes and throughput rises.
Every ParalleliX AI inference comes with a SHA-256 commitment over the request, result, and node ID. The proof lives on the response. You can verify it yourself. Centralized AIs cannot offer this.
request_id: req_a3f29b7c1e…
committed_hash: SHA-256(request_id · result · node_id · timestamp)
node_id: 3 · served_at: 2026-06-23T17:54:53Z
This badge appears on every ParalleliX AI response. No centralized AI can offer cryptographic proof of where your inference ran.
One command installs the CLI. Four steps from zero to earning. No approval needed. No slashing. 0% platform fees.
npm i -g @parallelix/node-cliOne command. Detects your hardware automatically.
parallelix-node setupCreates your node identity. Prints a staking link.
Stake $PRLX on ConsoleRegister on-chain. No slashing. Stake always returned.
parallelix-node startHeartbeats, inference serving, streaming $PRLX rewards.
Paid for uptime, not per request. 0% platform fees. Render charges 12%.
A node is a machine you stake to register and keep online. The coordination layer routes inference requests to it based on what your hardware can run, and pays you for uptime. ParalleliX AI is where most of that work comes from.
ParalleliX AI sends inference requests to the coordinator. The scheduler routes each to a capable, online node. Whole requests, not fragments. Add nodes and throughput rises.
Each result arrives with a cryptographic commitment over (request_id, result, node_id). You can verify which node served your request and that the result is unaltered.
Requests match nodes on three signals: capability fit, capacity-aware queuing, and best-fit allocation. Higher uptime earns a larger weighted pool share.
Open-source models served by the network's own operator nodes. Every response carries a verified Proof-of-Execution. Free to use, credits for unlimited access.
Open-source models only. No proprietary weights.
Every answer is served by a real GPU node and carries a verified Proof-of-Execution.
Deposit $PRLX into an on-chain credits vault. Spend per message, metered off-chain, no per-message gas. Withdraw unspent credits any time.
Funds live in an immutable on-chain vault, withdrawable any time. The off-chain meter debits usage, and the operator share settles on-chain per served request.
chat, embeddings, classification
parameter sweeps, genomics
map-reduce, risk, logistics
proof generation, oracles
frames, encoding, AR/VR
A developer API over the same nodes ParalleliX AI uses. Point an OpenAI-compatible client at it, fan a batch of prompts across the network in parallel, or wire it into Claude with the MCP connector. Calls spend $PRLX credits.
# ParalleliX Compute API
POST /v1/chat/completions OpenAI-compatible completion
POST /v1/batch fan prompts out across the network
GET /v1/batch/{id} per-item result + Proof-of-Execution
GET /v1/models models the network serves now
GET /v1/usage requests, credits spent, balance
# Authorization: Bearer pk_live_... key spends $PRLX credits
# 85% of every paid call settles to the serving operatorsPOST /v1/chat/completions in the OpenAI format. Point an existing client at ParalleliX by changing one base URL.
POST /v1/batch with an array of prompts. They fan out across online nodes at once, each returning its own result and Proof-of-Execution.
npx parallelix-mcp gives Claude a parallel_map tool. Your agent orchestrates; the open-source fleet runs the bulk parallel sub-tasks.
One key, one billing rail. Every call is metered off-chain against the key wallet's $PRLX credit balance, the same balance the chat app uses.
Open-source models, currently 7B-class. A cheap parallel executor for bulk independent sub-tasks, not a frontier model.
The $PRLX contract owner is the zero address. No admin keys. No pause. No mint. The 4% tax is frozen. Verified on Etherscan, permanent. Stake $PRLX to earn $ETH and $PRLX from the reward pool.
100M
Total Supply · Fixed
No inflation. No mint function. The supply is final.
4%
Tax · Frozen
Split across development, ops, marketing, and holder revenue. Permanent.
25M
Operator Rewards Pool
Streamed per-second to online nodes. Weighted by stake, tier, and uptime.
Token Distribution
$PRLX Contract · Renounced
0x93FF39f6…893