The 0G Galileo testnet is the public sandbox for development and testing. The SDK auto-detects testnet from the chain ID — pass network="testnet" to create_broker() to connect.

Network parameters

ParameterValue
Network name0G-Galileo-Testnet
Chain ID16602
Token symbol0G
Block explorerchainscan-galileo.0g.ai
RPC endpointhttps://evmrpc-testnet.0g.ai
The default RPC is suitable for development. For production traffic, use a third-party provider (QuickNode, ThirdWeb, Ankr, dRPC NodeCloud).

Compute contract addresses

The SDK ships these addresses and resolves them automatically from the chain ID. You only need them if you’re calling the contracts directly outside the SDK.
ContractAddress
Ledger0xE70830508dAc0A97e6c087c75f402f9Be669E406
Inference Serving0xa79F4c8311FF93C06b8CfB403690cc987c93F91E
Fine-tuning Serving0xC6C075D8039763C8f1EbE580be5ADdf2fd6941bA
Override them in code if needed:
broker = create_broker(
    private_key       = "0x...",
    network           = "testnet",
    ledger_address    = "0x...",
    inference_address = "0x...",
    fine_tuning_address = "0x...",
)

Connect from Python

from zerog_py_sdk import create_broker

broker = create_broker(
    private_key = "0xYOUR_PRIVATE_KEY",
    network     = "testnet",
)
print(f"Chain ID: {broker.web3.eth.chain_id}")   # 16602
print(f"Address:  {broker.get_address()}")

Get testnet tokens

SourceURLLimits
Official faucetfaucet.0g.ai0.1 0G per wallet daily
Hackathon faucet0g-faucet-hackathon.vercel.appUse code AGENT-2026 for additional tokens
Google Cloud faucetWeb3 faucet

Dev mode

Set ZG_DEV_MODE=true (or NEXT_PUBLIC_ZG_DEV_MODE=true) to use the testnet_dev contract addresses, which point at a separate development deployment on the same chain ID. Most users won’t need this.
export ZG_DEV_MODE=true

Add to MetaMask

FieldValue
Network name0G-Galileo-Testnet
New RPC URLhttps://evmrpc-testnet.0g.ai
Chain ID16602
Currency symbol0G
Block explorer URLhttps://chainscan-galileo.0g.ai

Next steps

Mainnet

When you’re ready for production traffic.

Inference

Run your first AI inference on testnet.