logo

Does Base Have a Native Token?

Base uses ETH as its gas token and does not have a separate native network token. Transactions on Base are paid for in ETH bridged to or acquired on the network, so an account holding only USDC cannot submit transactions until it also holds a small amount of ETH, unless a paymaster covers fees on its behalf.

What pays for a transaction

Base is a layer-2 network that settles to Ethereum, and it uses ETH as its gas currency. Submitting a transaction requires an ETH balance on Base to cover the fee, exactly as a transaction on Ethereum requires ETH there. There is no separate token that must be held or staked to use the network.

This matters operationally more than conceptually. A funded agent is often thought of as an agent with a USDC balance, because USDC is what it spends. But a token transfer is still a transaction, and the fee for it is paid in ETH. An account with a large USDC balance and no ETH cannot move anything. The failure appears at submission and looks like a broken integration rather than a funding problem, which is why it costs more debugging time than it should.

Fees on Base are generally small, which makes frequent machine payments practical in a way they are not on more expensive networks. Small is not zero, and a process making continuous payments consumes ETH continuously.

Claims of an official token

Because networks commonly do have their own tokens, people reasonably assume Base does, and that assumption is a well-worn lure. Tokens are regularly promoted as the official Base token, and none of them are. There is nothing to buy in order to use the network, no airdrop to claim in exchange for connecting a wallet or signing a message, and no staking requirement. Treat any prompt to acquire a Base token, or to sign something to claim one, as hostile by default. The general rule applies here with full force: verify a token by its contract address against the issuer's own published source, and be suspicious of any flow that arrives with urgency attached.

Keeping an autonomous account able to transact

Fund gas separately from spend. Treat the ETH balance as its own resource with its own floor. A design where the agent's working capital is USDC and its gas is a small ETH reserve makes the two failure modes distinct: out of money to spend, and out of money to transact.

Alert on the gas balance before it runs out. A threshold alert on ETH is one of the highest-value monitors for an autonomous payer, because the failure is total and silent from the agent's perspective. It simply stops being able to act.

Automate top-ups with a bound. Refilling the ETH reserve when it drops below a floor keeps the agent alive, and the refill should itself be capped and logged so a loop cannot drain a funding source.

Consider a paymaster where the account abstraction supports it. A paymaster can pay fees on behalf of an account, which removes the requirement for the agent to hold ETH at all and shifts it to whoever sponsors the paymaster. That is a real simplification and it introduces a dependency, so it is a trade rather than a free win.

Account for fee variability. Fees move with network conditions. Logic that assumes a fixed cost per transaction will be wrong sometimes, and an agent that budgets to the penny can find a transaction it cannot afford.

Watch for failed transactions consuming gas. A reverted transaction still costs fees. A retry loop against a consistently failing call can burn a reserve without accomplishing anything, which is why retries need both a cap and an alert.

Estimating what to hold

A reasonable approach is to measure rather than guess. Run the agent's real workload for a period, record the fee actually paid per transaction, and multiply by expected volume over the period between top-ups, then add headroom for fee variability and for retries. The result is usually a small number, which is the point: on a network with low fees the gas reserve is not a meaningful cost, and the reason to think about it at all is availability rather than expense. What you are buying with the headroom is the guarantee that the agent does not stop at an inconvenient moment, and that guarantee is worth considerably more than the ETH sitting idle to provide it.

Frequently asked questions

Does Base have its own token?
No. Base uses ETH as its gas token and has no separate network token. There is nothing to buy, stake, or claim in order to use the network, so any token promoted as the official Base token is not one and any claim flow asking you to sign for one should be treated as hostile.
What token do you pay gas in on Base?
ETH, held on Base itself. A transaction cannot be submitted without an ETH balance on the network to cover its fee, which is why an account holding only USDC will fail at submission even though it has plenty of value available to spend.
Can an agent hold only USDC?
Only if something else covers its fees. Otherwise it needs a small ETH balance on Base, because a token transfer is itself a transaction with a gas cost. A paymaster, where the account abstraction supports one, can pay those fees so the agent holds no ETH.
How much ETH does an agent need on Base?
Enough to cover its expected transaction volume with headroom for fee variability, kept above a floor by monitoring and automated top-ups. Fees on Base are typically small but they move with network conditions, so budgeting an exact fixed cost per transaction will occasionally be wrong.
Do failed transactions still cost gas?
Yes. A transaction that reverts has still consumed execution, so the fee is paid. This matters for autonomous agents because an unbounded retry loop against a consistently failing call can exhaust a gas reserve while accomplishing nothing, which argues for capped retries with alerting.