Until now, Starknet employed a forecast-based fee model. The price took into account future optimizations (like S-two) before they were technically live. This was done to be as cheap as feasible, but had a few bad consequences. First, this created a disconnect: major engineering breakthroughs launched without a corresponding drop in gas fees, because the discount was already baked in. This made the network’s technical progress invisible to users. Secondly, it created a situation where prices are losing sync with reality. If we projected 60% improvement and “only” improved by 50% - transactions would be underpriced (and vice versa).
We are now shifting to a real-time cost alignment model. We are adjusting gas fees to reflect the current real-world operational costs of a high-performance network. This means fees will be higher in the short term, yet still well within the industry baseline. (Current suggestion: go from 3gFri/L2gas to 8 gFri/L2gas). In any case, prices are expected to get lower with more optimizations, which users will feel in real time. In any case we will always make sure that transaction fees will at least cover their cost, creating a sustainable economic baseline.
We strongly believe that this change of mindset paves the way to sustainable decentralization of Starknet, which is one of our main goals. On top of that, this mindset opens the door to many potential features that may come here. Do we want to adapt the fee structure if the STRK price significantly changes? And how to decide what is “break even” when there are decentralized sequencers?
What do you think of this suggestion? What future improvements would you like to see in the fee market?
Is this change going to affect the rewards received by the validators ?
No, while its true that this suggestion will increase fees validators pay, we still talk about few cents per hour, which accumulates to much less than $100 a month. This is still far less than the infrastructure cost validators are expected to pay in the decentralization era. If this creates issues for you as a validator - it means it moves earlier a financial issue we would anyway have with full decentralization. Happy to disucss if this is the case.
My opinion:
Some fee increase may be justified, but the real issue is the lack of predictability. Today the USD cost of transactions changes only because STRK price moves - not because gas usage increased.
-
The “<$0.001 per simple tx” claim doesn’t reflect real DeFi usage: swaps cost ~$0.02, routed swaps often ~$0.05–0.08.
-
Since v0.14.0 we pay fees in STRK, but there’s no mechanism keeping them stable in USD. The same action doubled in price when STRK doubled. If STRK hits $2, will a basic tx cost ~$0.40? This is critical for high-volume users.
-
The proposed changes effectively push active users toward relying on wallet paymasters to hide the volatility.
Summary:
I support the idea of fee adjustments, but the network needs a more stable and predictable fee model. Users shouldn’t overpay when STRK rises, and sequencers shouldn’t under-earn when it drops. A balanced solution benefits everyone.
I think that there are two issues here, which are totally seperate - the way we presented the increase (and whether price is now too high) and future plans with STRK volatility. so I’ll split my answer to two
-
Current prices: I agree that the “<0.001$ for transfer” isn’t reflecting the whole picture (this is why you will find it only in the tweet, and not here, where I try to discuss the full implications). We are aware to many use-cases where a transaction indeed costs several cents now. Question is if after the increase any commonly used usecase will be prohibitively expensive
-
STRK affect: While I’m not sure its inherently bad that STRK goes up means fee goes up (this is what happens in any L1, practically. No? Am I missing something?) , I agree a linear correlation is too much given that the “basis point” is the transaction $$ cost, and maybe we want to have an upper ceiling there/square root/other mechanism. (also - a true - long term paying fees in $$ but nominate it in STRK is a bit counter-productive to the reason to use STRK in the first place)
We are currently disucssing internally what features we can/want to develop to somehow change this automatically and what formulas to use. For now - we are trying to be break-even with costs, and if STRK cost will go crazy for either direction we will modify fees accordingly (though now this is still a manual operation which might take a few days). Happy to get feedback on this line of thought, here or on a call
I want to add one more angle to the discussion.
If Starknet eventually moves away from Ethereum DA and becomes more independent, then a pure STRK-denominated fee market makes perfect sense. In that scenario, volatility isn’t a problem - STRK becomes the native unit of account, and the ecosystem is self-contained.
But as long as we are an L2, the expectations are different. The whole point of an L2 is to give users stable, cheap, predictable fees instead of paying volatile L1 prices. A direct 1:1 correlation between STRK price and tx fees breaks that expectation. Users end up paying more not because the network costs increased, but because the token pumped.
So unless the long-term vision truly is to detach from Ethereum and operate more like an L1, some smoothing or stabilization mechanism feels necessary. Otherwise volatility will keep pushing users toward paymasters just to regain predictable UX.
Another thing I’d like clarity on is the “break-even” logic. Right now we hear that fees need to cover operational costs - but it’s unclear what hardware the sequencer currently uses and what “optimal infrastructure” actually means. When decentralization arrives and we, as validators/sequencers, are supposed to run this infrastructure and receive rewards, it would help to understand what cost assumptions we should plan for.
Predictability matters not only for users today, but also for future operators.
Regarding (2), I agree with saniksin. My expectation is that transaction fees should be stable in USD terms and not be subject to STRK volatility. The promise of Starknet in my mind has always been to scale through increasing bandwidth rather than bidding up a fixed amount.
Transaction fees increasing due to STRK price increases puts a damper on network growth, since the relative cost of a new entrant is higher than and existing holder, and makes it hard for companies to build sustainable business models that incorporate paymaster etc since real world liabilities are USD denominated (or some other more stable base currency).