Fee Structure

We wanted fees to feel familiar to Ethereum developers while supporting the new workloads that an AI-native chain introduces. Citrate's fee structure extends the familiar base fee plus priority tip model with additional fee types for AI inference and bridge operations. Every fee on the network is denominated in SALT, and a portion of all fees is burned to create deflationary pressure on token supply.

Transaction Fees

Standard transaction fees on Citrate follow the EIP-1559 model used by Ethereum. Each transaction pays a base fee (determined by network congestion) plus an optional priority tip (to incentivize faster inclusion by miners).

Total Fee = (Base Fee + Priority Tip) x Gas Used

The base fee adjusts dynamically based on block utilization:

If the previous block was more than 50% full, the base fee increases by up to 12.5%
If the previous block was less than 50% full, the base fee decreases by up to 12.5%
The minimum base fee is 1 gwei (0.000000001 SALT)

# Check current base fee
cast basefee --rpc-url https://rpc.cnidarian.cloud

# Estimate gas for a transaction
cast estimate 0xTARGET_ADDRESS "transfer(address,uint256)" 0xRECIPIENT 1000000000000000000 --rpc-url https://rpc.cnidarian.cloud

Because Citrate uses GhostDAG with parallel block production, effective throughput is much higher than a single-chain blockchain. This means base fees tend to be lower than on Ethereum for equivalent demand levels, as congestion is distributed across parallel blocks.

AI Inference Fees

Inference fees are paid when a smart contract or user requests AI model execution through the InferenceEngine precompile. These fees are separate from gas fees and are priced by model operators.

The total inference fee has three components:

Component	Description	Typical Range
Model fee	Set by the model operator, reflects compute cost	0.001 - 1.0 SALT
Network fee	Protocol fee for routing and settlement	15% of model fee
Verification fee	Additional cost for higher verification tiers	0 - 0.5 SALT

Verification tier surcharges:

Tier	Surcharge	Why
Signature	0 SALT	Minimal overhead (operator signature only)
Optimistic	0.005 SALT	Covers potential dispute resolution costs
ZK-SNARK	0.05 - 0.5 SALT	Covers proof generation compute costs

// Estimate inference cost before submitting
function estimateCost(bytes32 modelId) public view returns (uint256) {
    (, , , uint256 price, ,) = IModelRegistry(address(0x0100)).getModel(modelId);
    uint256 networkFee = price * 15 / 100;
    return price + networkFee;
}

Bridge Fees

Cross-chain bridge operations incur fees that compensate oracle nodes for their attestation work and cover the gas costs on destination chains.

Operation	Fee	Breakdown
Deposit (external chain to Citrate)	0.1% of value	70% to oracles, 20% to treasury, 10% to LPs
Withdrawal (Citrate to external chain)	0.1% of value + destination gas	Same split plus destination chain gas
Fast withdrawal (skip confirmation wait)	0.3% of value	Premium for immediate processing

Bridge fees are deducted from the transferred amount. For example, bridging 1000 USDC to Citrate costs 1 USDC in bridge fees, and the user receives 999 USDC worth of wrapped tokens.

# Estimate bridge fee
citrate-cli bridge estimate --from-chain ethereum --to-chain citrate --amount 1000 --token USDC --rpc https://rpc.cnidarian.cloud

Fee Burn Mechanism

Citrate burns a portion of all fees, permanently removing SALT from the circulating supply. This creates a deflationary force that intensifies as network usage grows.

Fee Type	Burn Rate	Recipient of Remainder
Transaction base fee	50% burned	50% to block producer
Transaction priority tip	0% burned	100% to block producer
Inference model fee	15% burned	80% to operator, 5% to verifier
Inference network fee	100% burned	--
Bridge fee	10% burned	Distributed to oracles, treasury, LPs

The burn mechanism means that periods of high network activity directly reduce token supply. During peak usage, the daily burn rate can exceed the daily emission from mining, creating net deflationary periods.

# Check cumulative burn statistics
citrate-cli tokenomics burn-stats --rpc https://rpc.cnidarian.cloud

Example output:

Total burned:          12,345,678 SALT
24h burn rate:         15,234 SALT
24h emission rate:     12,500 SALT
Net 24h supply change: -2,734 SALT (deflationary)

Gas Costs for AI Operations

AI precompile operations have specific gas costs that differ from standard EVM opcodes:

Operation	Gas Cost	Notes
ModelRegistry.register	42,000 + storage	One-time registration cost
ModelRegistry.getModel	2,100	Equivalent to cold SLOAD
InferenceEngine.requestInference	21,000 base	Plus model-specific compute fee
InferenceEngine.fulfillInference	15,000 + calldata	Paid by the network, not the requester
AdapterManager.createAdapter	32,000 + weights	Depends on adapter size
TokenScorer.scoreAddress	5,000	Lightweight on-chain scoring
ContextBridge.storeContext	20,000 + storage	Depends on context size
AttestationVerifier.verify	15,000	Single attestation verification