Crypto Gloss

Gas Token

Gas tokens were a category of ERC-20 tokens — most notably GST2 (Gas Savings Token 2) and CHI (developed by 1inch) — that exploited Ethereum’s EVM storage refund mechanism to allow users to effectively purchase gas during periods of low network congestion and spend that pre-purchased gas later during high-congestion, high-price periods, by encoding gas “storage” into on-chain state slots that yield refunds when cleared, effectively arbitraging gas price differences across time in a way that dramatically reduced transaction costs for sophisticated DeFi users and protocols during the 2019–2021 period before EIP-3529 (London hard fork, August 2021) eliminated most storage refunds and made gas tokens functionally obsolete. Gas tokens were a clever but complex optimization that demonstrated both the incentive design depth of EVM economics and how unexpected economic behaviors emerge from precise fee rules.

The EVM Storage Refund Mechanism

Original Rule (Pre-EIP-3529)

“`

SSTORE (new slot, 0 → nonzero): costs 20,000 gas

SSTORE (delete slot, nonzero → 0): refund 15,000 gas

Maximum refund per transaction: gasUsed / 2

Implication:

If a transaction uses 100,000 gas and deletes many storage slots

→ receive up to 50,000 gas back as refund

→ effective transaction cost: 50,000 gas × gas price

“`

The refund created an economic incentive: “pre-buy” gas when cheap by writing storage, receive refund later during expensive transactions.

How Gas Tokens Worked

GST2 (Gas Savings Token 2) — by Project Chi, 2019

Minting (Banking Gas):

“`

Call GST2.mint(amount) when gas price is 1 gwei:

Each token minted: consumes ~35,000 gas writing to storage slots

Cost per token: 35,000 × 1 gwei = 0.000035 ETH ($0.006 at $200 ETH)

After minting 100 GST2:

Stored on-chain: 100 storage slots (3.5M gas “banked”)

Total cost: 0.0035 ETH (~$0.70)

“`

Burning (Spending Gas):

“`

When gas price is 200 gwei (high congestion):

Include GST2 burn in the same transaction:

100 tokens burned → deletes 100 storage slots

Refund: 100 × 15,000 = 1,500,000 gas refunded

Refund value: 1,500,000 × 200 gwei = 0.3 ETH

Without gas token: transaction costs 0.3 ETH in gas fees

With gas token: transaction costs 0.3 ETH – 0.3 ETH refund + 0.0035 ETH minting cost

Net savings: up to ~$60+ per complex transaction at 2021 gas prices

“`

CHI Token — by 1inch

1inch’s CHI token improved on GST2:

  • More efficient storage encoding: higher refund-per-token ratio
  • Integrated directly into 1inch aggregator — auto-burned on eligible swaps
  • CHI minting available at any time; 1inch UI showed CHI balance and savings

Gas Token Economics

When Profitable to Use

“`

Gas price at burn > Gas price at mint × (refund gas / mint gas)

Example:

Mint at 10 gwei (100 CHI, cost: ~$5)

Burn when gas > 30 gwei:

Refund > mint cost → net savings

At 100 gwei: savings of ~$50+ per complex DeFi transaction

“`

When Not Profitable

  • Transaction complexity too low (refund cap = gasUsed / 2)
  • EIP-3529 in effect (refund mechanism changed)

EIP-3529: The End of Gas Tokens

EIP-3529 (August 2021, London hard fork) made two critical changes:

Change 1: SELFDESTRUCT Refund Eliminated

Change 2: SSTORE Refund Reduced

  • Maximum refund per transaction: reduced from gasUsed / 2 to gasUsed / 5

Impact

Old refund per GST2 token burned: 15,000 gas

New refund per GST2 token burned: 4,800 gas

Original GST2 mint cost per token: ~35,000 gas

Breakeven: needed refund > mint cost at some future gas price

Old: 15,000 / 35,000 = 0.43× → profitable if future gas is 2.3× mint price

New: 4,800 / 35,000 = 0.14× → would need future gas to be 7× mint price for same benefit

Result: Gas tokens became economically unviable for most use cases

“`

Legacy Impact

While gas tokens as a product are obsolete, their design influenced DeFi development:

  1. Storage slot optimization is now a core Solidity best practice — understanding storage costs led directly to struct packing, immutables, and event-over-storage patterns
  2. Reflexive arbitrage mechanisms — the idea of buying a resource when cheap and using it when expensive influenced token economic design
  3. EIP-3529 itself was partially motivated by gas token abuse, which made Ethereum state size grow unsustainably

History

  • 2017: “Gas station” concept explored in academic/forum discussions
  • 2018: GST1 (Gas Savings Token 1) launched as proof of concept; limited adoption
  • 2019: GST2 launched with improved efficiency; first mainstream gas token
  • 2020: 1inch launches CHI token; integrated into 1inch aggregator; gas token use peaks among DeFi power users
  • 2020–2021: Gas token minting becomes common during low-gas periods (nights/weekends); CHI widely held by DeFi whales
  • 2021 Aug: EIP-3529 activated (London hard fork); GST2 and CHI refunds drop ~70%; gas tokens immediately become uneconomical for new minting
  • 2021–2025: Gas tokens fade into historical curiosity; existing holders use residual refund value; no new gas tokens emerge

See Also