Lux Docs

On-chain AI compute task orchestration, TEE attestation, and reward distribution.

The AI VM (A-Chain) handles AI compute task orchestration on the Lux network. Compute providers register their GPU resources, tasks are submitted and assigned to providers, and results are verified via Trusted Execution Environment (TEE) attestation. Rewards are distributed per epoch and state is periodically anchored to the Q-Chain via merkle roots.

Architecture

Provider Registry: GPU compute providers register with attestation proof
Task Queue: Tasks submitted, assigned to providers, results verified
TEE Attestation: CPU (SGX/SEV-SNP/TDX) and GPU (nvtrust) attestation -- local verification, no cloud dependency
Reward Distribution: Per-epoch rewards based on compute contribution
Merkle Anchoring: Periodic state root anchoring to Q-Chain for verifiability

Configuration

{
  "maxProvidersPerNode": 100,
  "maxTasksPerProvider": 10,
  "requireTEEAttestation": true,
  "minTrustScore": 50,
  "attestationTimeout": "30s",
  "maxTaskQueueSize": 1000,
  "taskTimeout": "5m",
  "baseReward": 1000000000,
  "epochDuration": "1h",
  "merkleAnchorFreq": 100
}

Key Parameters

Parameter	Default	Description
`maxProvidersPerNode`	100	Maximum providers per node
`maxTasksPerProvider`	10	Maximum concurrent tasks per provider
`requireTEEAttestation`	true	Require hardware attestation for providers
`minTrustScore`	50	Minimum trust score (0-255) to accept tasks
`taskTimeout`	5m	Maximum time for a task to complete
`baseReward`	1 LUX	Base reward per epoch (in wei)
`epochDuration`	1h	Duration of each reward epoch
`merkleAnchorFreq`	100	Blocks between Q-Chain merkle anchors

REST API

The A-Chain uses a REST API (not JSON-RPC). Endpoint:

https://api.lux.network/{network}/ext/bc/{blockchain-id}

Provider Endpoints

GET /providers -- List all registered compute providers
POST /providers/register -- Register a new compute provider with GPU info and attestation

Task Endpoints

GET /tasks -- List pending tasks (or get a specific task by ?id=...)
POST /tasks/submit -- Submit a new AI compute task
POST /tasks/result -- Submit task result from a provider

Model Endpoints

GET /models -- List available AI models

Attestation Endpoints

POST /attestation/verify -- Verify a TEE attestation proof

Reward Endpoints

POST /rewards/claim -- Claim accumulated rewards
GET /rewards/stats -- Get reward statistics

Status Endpoints

GET /stats -- Get A-Chain statistics (providers, tasks, throughput)
GET /merkle -- Get current merkle root
GET /health -- Health check

Example: Register a Provider

curl -X POST \
  https://api.lux.network/mainnet/ext/bc/{blockchain-id}/providers/register \
  -H "Content-Type: application/json" \
  -d '{
    "id": "provider-1",
    "wallet_address": "0x...",
    "endpoint": "https://gpu.example.com",
    "gpus": [
      {
        "model": "A100",
        "memory": 80000,
        "compute_capability": "8.0"
      }
    ],
    "gpu_attestation": {
      "type": "nvtrust",
      "evidence": "..."
    }
  }'

See lps.lux.network for the relevant Lux Proposals governing this chain.

AI VM (A-Chain)

On this page