What made network fees so low?

[Alpha Marine]

I know it's a noob like question, but I've been struggling with it.
For about a year or two now, Bitcoin's transaction fees have been extremely low. I remember times when we had to deal with congestion because of ordinals, BRC-20 and the rest, and now it seems like that is a thing of the past because for a very long time we haven't had that issue. I would simply like to know what happened. Were there upgrades in the new versions of Bitcoin Core to prevent things like this, or are they still in the network but can't cause congestion?

Secondly, even when there is no network congestion, in the past, network fees were not this low. Currently, the fee for a high-priority transaction is fluctuating between 1-3 sats/vB, which is about $0.30. In the past, in normal networks, the fees could go about $1+ or closer to $2, so what changed in the last two years?

And I don't believe it is the "lower network congestion or less competition for block space" answer I get whenever I try to read up about it. I'm not convinced by that answer because when I look at the daily trading volume of Bitcoin, it has even increased, and the average daily transactions have increased over the years too. This means there are evidently more transactions now than two or three years ago. This means there should be more congestion now, so why aren't there any?

[Upgrade00]

I remember times when we had to deal with congestion because of ordinals, BRC-20 and the rest, and now it seems like that is a thing of the past because for a very long time we haven't had that issue. I would simply like to know what happened. Were there upgrades in the new versions of Bitcoin Core to prevent things like this, or are they still in the network but can't cause congestion?

The congestion caused by ordinals and runes was a hype at the time, it was always going to fade away and free up the space on the network. On both sides of that period the network fee was within the normal range it is in now.

Secondly, even when there is no network congestion, in the past, network fees were not this low. Currently, the fee for a high-priority transaction is fluctuating between 1-3 sats/vB, which is about $0.30. In the past, in normal networks, the fees could go about $1+ or closer to $2, so what changed in the last two years?

There were periods in the past that the fee rate was fluctuating within the same range it is in now. If you monitor the thread about consolidating your addresses, you will be able to see the different fee rate ranges that we had.

[Zaguru12]

I will say it like Satoshi, sorry I can’t convince you but let me try

For about a year or two now, Bitcoin's transaction fees have been extremely low. I remember times when we had to deal with congestion because of ordinals, BRC-20 and the rest, and now it seems like that is a thing of the past because for a very long time we haven't had that issue. I would simply like to know what happened. Were there upgrades in the new versions of Bitcoin Core to prevent things like this, or are they still in the network but can't cause congestion?

Bitcoin transactions have been low because there less network congestion and this is the outright truth, when there was congestion caused by ordinals and inscriptions, yes they have not total die down but the hype around them has definitely died down because like some scam proposals like ICOs and NFTs people do not gamble on them anymore.

There was never any consensus rule changes because of them, neither where there any block size increments or soft forks proposals like BIP-110 which were to temporarily to segregated this arbitrary data on bitcoin transactions were approved the consensus rule remains the same, BRC20 are not hidden somewhere.

Secondly, even when there is no network congestion, in the past, network fees were not this low. Currently, the fee for a high-priority transaction is fluctuating between 1-3 sats/vB, which is about $0.30. In the past, in normal networks, the fees could go about $1+ or closer to $2, so what changed in the last two years?

Let’s say there are many bitcoin users who uses addresses like Segwit and taproot addresses more now than in the early days of this soft fork were bitcoin legacy addresses were a thing, we know the difference that transactions between Segwit addresses and legacy addresses brings in terms of block size congestion. So this might have been the clear difference between now and then

Also there are many users of second layer networks now which takes some transactions temporarily off the base chain transactions to reduce network congestion

And I don't believe it is the "lower network congestion or less competition for block space" answer I get whenever I try to read up about it. I'm not convinced by that answer because when I look at the daily trading volume of Bitcoin, it has even increased, and the average daily transactions have increased over the years too. This means there are evidently more transactions now than two or three years ago. This means there should be more congestion now, so why aren't there any?

You don’t need to look only at trading volume because there are even offline or internal transactions on exchanges which contributes to the bitcoin trading volume and they do not contribute to onchain transactions example is the usual future or spot trading on exchanges

[Cookdata]

I know it's a noob like question, but I've been struggling with it.
For about a year or two now, Bitcoin's transaction fees have been extremely low. I remember times when we had to deal with congestion because of ordinals, BRC-20 and the rest, and now it seems like that is a thing of the past because for a very long time we haven't had that issue. I would simply like to know what happened. Were there upgrades in the new versions of Bitcoin Core to prevent things like this, or are they still in the network but can't cause congestion?

There was a Bitcoin Core upgrade, but it wasn't what reduced the fee drastically. Ordinals inscriptions came with hype, some came as airdrops, and because everyone wanted a piece of the cake, the trend caused mempool disturbance and spam, but as the hype dies down, it wasn't too profitable, so the activity stopped, and the mempools became free. You can see from the chart below how the fees reduced over the years.



https://dune.com/dataalways/ordinals

The Bitcoin Core upgrade that happened was to encourage people to use Op_return to include arbitrary data instead of exploiting taproot with ordinals. This was last year between September and October when Bitcoin Core version 30 was released with data carrier size up to 100kb instead of the old versions with 83 bytes.

Secondly, even when there is no network congestion, in the past, network fees were not this low. Currently, the fee for a high-priority transaction is fluctuating between 1-3 sats/vB, which is about $0.30. In the past, in normal networks, the fees could go about $1+ or closer to $2, so what changed in the last two years?

And I don't believe it is the "lower network congestion or less competition for block space" answer I get whenever I try to read up about it. I'm not convinced by that answer because when I look at the daily trading volume of Bitcoin, it has even increased, and the average daily transactions have increased over the years too. This means there are evidently more transactions now than two or three years ago. This means there should be more congestion now, so why aren't there any?

I think it's all about block demand. When everyone fights to get their transaction fast in the next block, it increases the fees naturally, but since ordinals went to an all time low, there was enough space for everyone to get their transaction confirmed. As the mempool became stable with time, people no longer overpaid fees. At some points, some nodes reduced their minimum relay fee to accept transactions that are below 1 sats/vbyte and gradually some other nodes and miners did the same, but not all miners and nodes have reduced their minimum relay fee because of spam.

[NotATether]

The default minimum fee allowed to be relayed by Bitcoin Core for transactions was slashed from 1 sat/vbyte to 0.1 sat/vbyte several months ago.

This basically collapsed fees by 10x overnight, when they were already at 1 sat/vbyte before.

[Mia Chloe]

~snip

Well in simple terms it's just the forks we implement. Basically it first started from the era when only legacy addresses that was when congestion was the biggest of problems then we got SEGWIT. But the thing is there are basically different versions. lately we got an even better upgrade which was taproot.

Basically taproot reduces fees by using the merkel tree to reveal only the executed spending path instead of the entire script which actually reduces the transaction size in bytes lowering the fees since Bitcoin fees are based on transaction weight (vbytes). So the more wallets allow taproot the more efficient transaction will be and the cheaper fees will be.

[Zaguru12]

Basically taproot reduces fees by using the merkel tree to reveal only the executed spending path instead of the entire script which actually reduces the transaction size in bytes lowering the fees since Bitcoin fees are based on transaction weight (vbytes). So the more wallets allow taproot the more efficient transaction will be and the cheaper fees will be.

Agree with you that the any of Segwit addresses either native Segwit (version 0) or Taproot (version 1) actually save fees. The merkle tree you mentioned though is actually a privacy advantage and then making multi sig transactions looking looking like a single sig transaction.

But the issue of saving fees or cheaper fees, it’s not in all transactions that Taproot is the best option, the native Segwit (bc1q) transactions are best for same input and output Transactions like 1 address to 1 address transactions or if the transaction has a smaller inputs than the outputs like spending from one input address to two output addresses (1 recipients and also a change address) or like signature payment from the campaign manager to multiple recipients addresses, the native Segwit is the best option because native Segwits has less output vbytes than taproot. A native Segwit address sending in 1 address to 1 address is around 141 vbytes but that of taproot is higher

But for a multiple input to few output transactions, taproot edges and saves fees example is the consolidation of little UTXOs to one this is because taproot addresses have smaller outputs vbyte compared to native Segwits

Then the multi sig transaction advantage while using taproot primarily due to the use of Schnorr signatures by the taproot addresses.

[Mia Chloe]

~snip

Even if lower fees generally is a good thing for the network I've always had this theory that if feeds continue to drop constantly and considerably like it already is especially after the drop in relay fee Notatether mentioned it might take us to a point where layer 2 seems less useful to the average spender.

Almost everyone using the lightning network at some point is mainly doing it to cut down fees because just a small percentage might be using it for privacy related reasons.

[Zaguru12]

Even if lower fees generally is a good thing for the network I've always had this theory that if feeds continue to drop constantly and considerably like it already is especially after the drop in relay fee Notatether mentioned it might take us to a point where layer 2 seems less useful to the average spender.

Almost everyone using the lightning network at some point is mainly doing it to cut down fees because just a small percentage might be using it for privacy related reasons.

Yes this is definitely right if the fees remain like this which is as a result of less network congestion I think Layer 2 usage for average users will definitely reduce. But I don’t think it will actually fully take away layer 2 because most of the layer 2 usage have secondary use case.

The secondary usage is fast transaction confirmation, I think if bitcoin actually gets adopted more and people actually starts to use bitcoin for payment methods like businesses accepting bitcoin payments, layer 2 will be considered for this micro payment and for fast transaction confirmation.

[nc50lc]

-snip-
And I don't believe it is the "lower network congestion or less competition for block space" answer I get whenever I try to read up about it. I'm not convinced by that answer because when I look at the daily trading volume of Bitcoin, it has even increased, and the average daily transactions have increased over the years too. This means there are evidently more transactions now than two or three years ago. This means there should be more congestion now, so why aren't there any?

Firstly, SegWit practically increased the block size past 1MB, now in "block weight" specifically up to 4M WU (weight units).

Secondly, those "less competition for block space" reasons that you've read read elsewhere is still applicable in a way.
Because the block space isn't based from how many transactions in it, but the weight of each transactions.

People mentioned ordinals but I see that you'll not take it for an answer because of your statement above.
But in the past two to three years, there could be lower volume of transactions than it is now (as you mentioned), but most of those are likely ordinals which are heavy in weight.
So overall, each full block can only contain less transactions that what it is today when there's less ordinals.

Lastly, as others have mentioned, the lower overall fee rate when there's no congestion is the now lower standard DEFAULT_MIN_RELAY_TX_FEE of 0.1sat/vB.

[bitbollo]

I would be not really optimistic about network fees... using onchain transaction for small UTXO would be relative more expensive in USD or FIAT Terms even in the future.
Not a real change for a "western" but much different from people living in a "poor economy".
Fees, in general, can become really expensive in a short time and leave expensive useless to use btc for an uncertain amount of time (at least for a certain amount, imagine micropayments).
About these micropayments,nowadays there are more "competitors" for this kind of payment and even systems that allow those to use secondary chains.
...be ready for block 999900 and so on... up to million and a few more blocks, the txfees would surpass easily 30-40 sats/byte

[Alpha Marine]

I appreciate the answers. I think it's clear to me now.

The default minimum fee allowed to be relayed by Bitcoin Core for transactions was slashed from 1 sat/vbyte to 0.1 sat/vbyte several months ago.

This basically collapsed fees by 10x overnight, when they were already at 1 sat/vbyte before.

Thanks. I had no idea about the default minimum fee reduction. I guess it makes sense now.

But in the past two to three years, there could be lower volume of transactions than it is now (as you mentioned), but most of those are likely ordinals which are heavy in weight.
So overall, each full block can only contain less transactions that what it is today when there's less ordinals.

Yeah, Upgrade00 already explained that ordinals were just hype and the hype died down.
I understood that there were fewer ordinals in the network, but I didn't understand why. My theory was that something was done to remove them from the network. I was wrong about that. Glad the hype died down.

Even if lower fees generally is a good thing for the network I've always had this theory that if feeds continue to drop constantly and considerably like it already is especially after the drop in relay fee Notatether mentioned it might take us to a point where layer 2 seems less useful to the average spender.

It will still be useful for fast transactions. No matter how low the fees get, most times, you still have to wait for 9-10 minutes. So I think it will always be useful.

[d5000]

The default minimum fee allowed to be relayed by Bitcoin Core for transactions was slashed from 1 sat/vbyte to 0.1 sat/vbyte several months ago.

This is indeed the best reason for the drastic fee collapse in a few months. However, it should be noted that some miners were accepting transactions with fees of less than 1 sat/vbyte even before that change. This is not a consensus setting but mempool policy. But to be relayed to the miners, users broadcasting transactions with low fee would need to find peers with low settings or connect directly to the mining pool's node. This was made drastically easier after that change, so it explains the collapse.

I'd like to add however another fact: 0.5 sat/vbyte at $60,000 per Bitcoin (a typical minimum fee as of my last checks in block explorers) is a similar fee than 5 sat/vbyte at a $6,000 Bitcoin, or 50 sat/vbyte at $600. As far as I remember, 50 sat/Byte (vbytes didn't exist back then) was indeed a typical fee in 2016 when Bitcoin was at $600, and in 2019/20, when Bitcoin was between $5000 and $10,000, you even got sometimes for 1 sat/vbyte, which was even much lower than now in USD. So yes, the fees are quite low now, but they were already lower in the past, only the Bitcoin price has increased.

It is also likely that the low fees will soon be part of the past: this chart shows that transaction activity is again close to record levels:


Source: https://bitinfocharts.com/comparison/bitcoin-transactions.html#alltime

[internetional]

The default minimum fee allowed to be relayed by Bitcoin Core for transactions was slashed from 1 sat/vbyte to 0.1 sat/vbyte several months ago.

This is indeed the best reason for the drastic fee collapse in a few months.

Aren't cause and effect reversed here? As far as I remember, fees dropped to 1 sat/vB first, and only later was the default parameter setting the minimum relay fee in Bitcoin Core lowered. So, in my view, lowering this parameter was a reaction to an objective drop in fees, rather than the factor that caused it.

Even if lower fees generally is a good thing for the network I've always had this theory that if feeds continue to drop constantly and considerably like it already is especially after the drop in relay fee Notatether mentioned it might take us to a point where layer 2 seems less useful to the average spender.

As for L2s - I've noticed that fees in some of them have increased significantly. For example, right now I don't even know how to transfer bitcoin from Spark via Lightning with a fee under 0.4%. Previously, you could do transfers with typical Lightning fees (which were much lower). Transfers from Bark via Lightning can even cost 0.5% of the total amount. Not long ago, this would have seemed crazy, but now it's the norm.

[Alpha Marine]

It is also likely that the low fees will soon be part of the past: this chart shows that transaction activity is again close to record levels:

But there are certain things we have to consider, aren't there?
The congestion was caused by the Runes token standard launch on the network. Without it, would there still have been network congestion and high fees?
From the chart, in April 2024, when transaction activity peaked, transaction fees had already started going down. It was around $30-$40 on that day, but it had already dropped from around $100. That shows that transaction activity can get that high and not experience a spike in fees.

Also, in April this year, the transaction activity went up to just over 830k a day, and from then till now, it has been moderately high, but we have not experienced any ridiculous increase in the fees. It even went as far as 890k in July, yet there were no issues, even though the record is just over 926k. So I doubt the current increase in transaction activity would change anything.

[d5000]

Aren't cause and effect reversed here? As far as I remember, fees dropped to 1 sat/vB first, and only later was the default parameter setting the minimum relay fee in Bitcoin Core lowered.

Yes and no at the same time As I wrote in my last post, there were miners already accepting sub-1 sat/vByte transactions before the change. But as it was still not the default value, the values didn't drop that much below the 1-sat value. The big drop, as far as I remember, came when Core changed the default setting, because then many intermediate nodes changed to the new value too, and it was made much easier to set that value in wallet software (the 1 sat/vByte setting was enforced several months more in some third party wallets).

Obviously you're correct that this was a reaction to the already lowering fees; default policy in Core is usually changed when the existing default doesn't align anymore with the "de facto" standard in the network (see also the OP_RETURN default which was changed in Core 30 as a reaction to it continually being "bypassed"). So if fees kept lowering, then the default value will probably be lowered again (e.g. to 0.01 sat/vByte), even if it is possible that (if Bitcoin's price doesn't explode to a million or more) there would be also resistance.

So the cause for the lowered default value was indeed the lowering fees, but it had also the effect of lowering the fees even further.

The congestion was caused by the Runes token standard launch on the network. Without it, would there still have been network congestion and high fees?
From the chart, in April 2024, when transaction activity peaked, transaction fees had already started going down. It was around $30-$40 on that day, but it had already dropped from around $100. That shows that transaction activity can get that high and not experience a spike in fees.

Also, in April this year, the transaction activity went up to just over 830k a day, and from then till now, it has been moderately high, but we have not experienced any ridiculous increase in the fees. It even went as far as 890k in July, yet there were no issues, even though the record is just over 926k. So I doubt the current increase in transaction activity would change anything.

Well the blocks are still not 100% full, and that's why the fees still are low. But they have already increased a bit. I remember some months ago in early 2026 the minimum fees were mostly around 0.2-0.4 sat/vbyte. Now they're almost consistently over 0.4 and often again above 1 sat/vByte, except on weekends. That's almost doubling.

The big spikes will occur if the blocks become full again for long stretches of time. I guess currently many Runes/Ordinals transactions are only created in periods of very low fees (<1 sat/vByte), and this means that there may also be a ceiling of the fees for a longer time if "monetary" transaction volume doesn't increase significantly. But judging from the numbers in the OP_RETURN observer, a big part of the transaction volume increase should not be related to Runes/Ordinals but indeed monetary, because while Runes activity has increased again, the magnitude of the general volume increase is much higher, and Ordinals are almost dead.

I guess that if we see consistently fees over 1 sat/vByte, Runes activity will again decline, but eventually the "monetary" transaction increase could make fees increase again. Maybe not to the "ridiculous" 2023/24 levels but to around 5 sat/vByte which would he also a quite healthy fee level I think - enough for the miners to survive long term if it persists, not too high to drive on-chain txes totally away, but also high enough to drive data transactions again to altcoins and tokens to L2s and other chains.

[ABCbits]

Aren't cause and effect reversed here? As far as I remember, fees dropped to 1 sat/vB first, and only later was the default parameter setting the minimum relay fee in Bitcoin Core lowered.

Yes and no at the same time As I wrote in my last post, there were miners already accepting sub-1 sat/vByte transactions before the change. But as it was still not the default value, the values didn't drop that much below the 1-sat value. The big drop, as far as I remember, came when Core changed the default setting, because then many intermediate nodes changed to the new value too, and it was made much easier to set that value in wallet software (the 1 sat/vByte setting was enforced several months more in some third party wallets).
--snip--

On the other hand, there are also mining pool that still/used to refuse to include TX with fee rate below 1 sat/vB. See https://bitcointalk.org/index.php?topic=5565739.msg66063663#msg66063663.

But judging from the numbers in the OP_RETURN observer, a big part of the transaction volume increase should not be related to Runes/Ordinals but indeed monetary, because while Runes activity has increased again, the magnitude of the general volume increase is much higher, and Ordinals are almost dead.

And i notice more ordinal TX have very small size. For example, https://mempool.space/tx/f811c4dd4c0bddcc6055b2a8f215de97b61f944a9f7cd5292d6081c5bfa5d976?showDetails=true shows a Rune/Runestone with 1 input and 2 output, but the TX size is only 94 vB. If we replace all of those with standard P2WPKH (native segwit), it would be about 140.5 vB instead.

[d5000]

And i notice more ordinal TX have very small size. For example, https://mempool.space/tx/f811c4dd4c0bddcc6055b2a8f215de97b61f944a9f7cd5292d6081c5bfa5d976?showDetails=true shows a Rune/Runestone with 1 input and 2 output, but the TX size is only 94 vB. If we replace all of those with standard P2WPKH (native segwit), it would be about 140.5 vB instead.

That's really interesting, do you know how these txes work or is this discussed somewhere here? Above all the address bc1pfeessrawgf which is used in the whole transaction sequence (and in a million of other transactions or so) seems strange. It seems to be an anyone-can-spend script. I have asked Gemini but the chatbot seems as clueless as I am about the purpose of these transactions, they seem to have to do something with the internal structure of Runes, but there is no true address in a large stretch of this chain of transactions.

Is this maybe a way to "register" a collection of Runes in the system with minimal footprint, so you can say you're the first? But that would contradict with the fact that the OP_RETURN string (which should "identify" the Rune) in the whole chain is the same, which seems like waste to me.

Anyway I think these are not "normal" Runes transactions because someone has eventually to own these "runes" on a normal private key and the transactions that really transfer ownership should be bigger than the standard P2WPKH transaction.


BTW, perhaps a better way to look at the possible fee evolution is the size of blocks in vBytes (this is not the true size but the size which includes the "witness discounts"). Basically 1 MB on this scale afaik means that "blocks are full".



The value is now around 780 kB. In 2025 the value was mostly around or below 700 kB, while in 2023/24 we had even higher values (up to ~850 kB). The high values before 2022 aren't really relevant becuase legacy (non-Segwit/P2PKH) scripts were still used a lot back then, which of course fill more virtual space in the blocks.

My guess is thus if this value approaches or surpasses 800 kB, then we will see again significantly higher fees.