Whats the current state of testnet3 vs testnet4?

[takuma sato]

I need to do some testing and I would like to know what is recommended to use right now, testnet3 or testnet4? Whenever I've asked this I remember people saying, do not use testnet4, it's very experiment, they are doing fringe things on it, and so on, and said testnet3 is more reliable.

Then other's said, do not use tesnet3, it's deprecated, and soon Bitcoin Core will stop supporting it, so you are wasting your time downloading and maintaining it.

I like to maintain a full blockchain for testnet because pruned mode do not allow for some tests. Testnet4 is last time I checked, substantially smaller, so that would save space.

Getting coins is also important. Id prefer availibility on faucets than someone here handling you coins for privacy.

So what would you recommend right now, testnet3 or testnet4? And please, mention which websites have good liquidty to some coins. Last time I check, I was only able to find some on coinfaucet, but would be good to have some alternatives in case they run out.

[OmegaStarScream]

It might be a good idea to just use the latest network. This is a good article[1] you may want to look into if you want to get more familiar with the differences between the different test networks. For the faucets:

- Mempool.space does support testnet4 as mentioned in the article[2], but I'm not sure which criteria one must meet to be able to use it. Simply making an account would just return "Your account does not allow you to access the faucet".

- Altquick[2]'s faucet supports testnet4, 3 and Signet[3].

- Another one I found is this[4]. It's not working at the moment (and not sure since when) but it might be worth keeping an eye on.

Alternatively, if you need some large amounts, you may want to make a request on the forums, or Reddit. Maybe someone will be willing to help and give you some.

[1] https://www.spark.money/tools/bitcoin-testnet-faucet-guide
[2] https://mempool.space/testnet4/faucet
[3] https://altquick.com/faucet/?q=TBTC4
[4] https://faucet.testnet4.dev/

[takuma sato]

It might be a good idea to just use the latest network. This is a good article[1] you may want to look into if you want to get more familiar with the differences between the different test networks.

Mempool.space does support testnet4 as mentioned in the article, but I'm not sure which criteria one must meet to be able to use it. Simply making an account would just return "Your account does not allow you to access the faucet".

Altquick[2]'s faucet supports testnet4, 3 and Signet[2].

Another one I found is this[3]. It's not working at the moment (and not sure since when) but it might be worth keeping an eye on.

Alternatively, if you need some large amounts, you may want to make a request on the forums, or Reddit. Maybe someone will be willing to help and give you some.

[1] https://www.spark.money/tools/bitcoin-testnet-faucet-guide
[2] https://mempool.space/testnet4/faucet
[3] https://altquick.com/faucet/?q=TBTC4

Thanks, I will give testnet4 a try. Has anyone synced a testnet4 full node without prune and assumevalid enabled? How big is your blocks folder? Also, did it took a lot of I/O work to sync? because I remember reading something along the lines of, "testnet4 is used for experiments that bloat this procedure a lot because of things like ordinals-type of scripts and so on". Not sure if that is true. I guess im going to try to sync it myself and see what happens.I already had problems syncing the regular blockchain on my laptop and had to use my desktop computer because it would run out of RAM, and it required several flushing events. Even on the desktop it happened a couple of times. I made a thread about this here:

https://bitcointalk.org/index.php?topic=5592968.msg67102751#msg67102751

It would be great if Bitcoin was only used for regular transactions to avoid these things but what can you do at this point. Most people don't care or realize that the number of laptops that can run open source BIOS is limited and often old hardware and non-monetary usages are pricing these machines out of the network.

[LoyceV]

I need to do some testing and I would like to know what is recommended to use right now, testnet3 or testnet4?

Both have their flaws: 3 has regular blockstorms, 4 has reorgs and empty blocks.

Id prefer availibility on faucets than someone here handling you coins for privacy.

If you want your test transactions to be private, you probably shouldn't use a blockchain

Thanks, I will give testnet4 a try. Has anyone synced a testnet4 full node without prune and assumevalid enabled? How big is your blocks folder? Also, did it took a lot of I/O work to sync?

It's 3 GB blocks, and 1 GB chainstate. Nothing to worry about.
Sorry, see below. It turns out mine was pruned.

PM me an address and I'll send you a coin (testnet4). Or 100. Doesn't matter

[nc50lc]

-snip- Has anyone synced a testnet4 full node without prune and assumevalid enabled? How big is your blocks folder? Also, did it took a lot of I/O work to sync?

My testnet4 blockchain is at 13GB right now, actual blocks directory size is at 12.7GiB.

getblockchaininfo:

Code:

{
  "chain": "testnet4",
  "blocks": 155027,
  "headers": 155027,
  "bestblockhash": "0000000000c6069cf027a59f9be1b9c6220c24c7602b2b8ef0f7a5c6ada07810",
  "bits": "1d00ffff",
  "target": "00000000ffff0000000000000000000000000000000000000000000000000000",
  "difficulty": 1,
  "time": 1791089788,
  "mediantime": 1791083783,
  "verificationprogress": 1,
  "initialblockdownload": false,
  "chainwork": "000000000000000000000000000000000000000000000ea85300d8c2d362ae06",
  "size_on_disk": 13641748547,
  "pruned": false,
  "warnings": [
  ]
}

I didn't disable -assumevalid but AFAIK, it wont affect your IO that much.
CPU on the other hand, will need more resources than usual.

For periodical cache flushes, 1GB of UTXO set that'll be written to disk roughly every hour isn't going to be as heavy as mainnet.

[DaveF]

Also, keep in mind you can use regtest:
https://developer.bitcoin.org/examples/testing.html

Bot 99% of testing it works just as well if not better then testnet.

Once you have it working there, THEN you bring it to testnet (if you think it's even needed).

Then you deploy.

Think of regtest and alpha and beta testing, testnet as RC testing and then live.

Just my view.

-Dave

[athanred]

Whats the current state of testnet3 vs testnet4?

All three testnets are altcoins, and they are traded for real BTCs:

https://altquick.com/market/bitcoin-testnet3-bitcoin/
https://altquick.com/market/bitcoin-testnet4-bitcoin/
https://altquick.com/market/bitcoin-signet-bitcoin/

So what would you recommend right now, testnet3 or testnet4?

Regtest, because they probably won't list a network with paper-and-pencil difficulty. Or a custom signet. Everything else is now an altcoin, so if you use it for real testing, then you risk touching real funds, and that could have some unintended consequences.

It might be a good idea to just use the latest network.

Then use testnet5. As long as they didn't list it yet, it should be good enough: https://github.com/bitcoin/bitcoin/pull/35861

Has anyone synced a testnet4 full node without prune and assumevalid enabled?

Not only that. People mined hundreds of thousands of tBTCs in testnet4 on their Windows 98 CPUs, and already sold a lot of them for real BTCs, when that coin was listed on exchange, because then it stopped being useful for testing. Now, you can use testnet5. Then, they will list it at some point in time, and then developers will sell everything, and move to testnet6. And so on.

It is sad to see testnets being turned into pump-and-dump altcoins, but well: if exchanges want to list them, then developers can keep launching new coins, and dumping the old ones. At least it has a side effect of funding the development, but it is just a consequence of altcoins being in such a bad state, that testnets, designed to be worthless, are worth more in practice, than some other serious altcoins, only because of testnets being included into Bitcoin Core.

testnet4 is used for experiments that bloat this procedure a lot because of things like ordinals-type of scripts and so on

It doesn't really matter. If anything, data pushes only make nodes faster, because OP_CHECKSIG, and processing real payments is more computationally expensive. Practically speaking, signet had some tests of hard-to-handle blocks, related to BIP-54, but testnet5 fixes it from the start, and signet tests were not permanent anyway, because slow blocks were reorged.

I already had problems syncing the regular blockchain on my laptop

Syncing the mainchain is much more difficult. I am at 85%, maybe I will sync it before the end of October, if I will be lucky. I started in June, and I am slowly getting there.

It is much easier to sync any testnet, because these chains are much smaller, and contain much less signatures, or other time-consuming scripts.

It would be great if Bitcoin was only used for regular transactions to avoid these things

If we would have the chain, where P2PK would be the only allowed address type, then it would be much harder to sync, and we would still have JPEGs.

the number of laptops that can run open source BIOS is limited

Of course. But some people, who claim running a node is easy, will tell you to trust, that UEFI is safe, and you should just get a better hardware, without caring about your BIOS being Open Source or not. Anyway, maybe they are right, because there is more code, that is in use, than you could read in a lifetime.

[nobswebdev]

I am using testnet4 for my project and it sometimes take long for tx to get confirmed. I don't have comparison to testnet3, because I never used it.

[LoyceV]

It is sad to see testnets being turned into pump-and-dump altcoins, but well: if exchanges want to list them, then developers can keep launching new coins, and dumping the old ones.

I'd say the testnet resets take far too long.

What would it take for me to create my own testnet4 chain fork? Setup a node and invalidate block #2 is enough, right? That should make it work for testing, mine coins regularly, and it's very easy to reset and start over by also invalidating my own block #2.
Anyone who invalidates block #2 can join my new, shorter chain.

[athanred]

What would it take for me to create my own testnet4 chain?

For testnet5, everything is ready, it is only a matter of launching it, and mining the Genesis Block. It is not yet deployed, because simply nobody did it yet.

Setup a node and invalidate block #2 is enough, right?

Even less than that: you don't have to modify Bitcoin Core at all, to reach a new testnet. Just setting up a mainnet node, and launching a mining pool is enough, to have a testnet, if you just save your partially-mined blocks. Then, you can test real mining, on real network, while also having a chance to hit a mainnet block once per lifetime or so. I think all testnets should work like that, but well, some people are against Merged Mining, so we won't have these things. And instead, they will keep complaining about mining being "in a terrible state", but they won't think about providing any incentive for home miners to produce any blocks.

I'd say the testnet resets take far too long.

Of course, because testnets should be resetted every 10 minutes, when the mainnet creates a new block. That would discourage exchanges from listing it, but well, people don't like that idea either.

Basically, if you would recompile mainnet code to allow weak blocks, then you would have your own testnet, resetted every 10 minutes. Which usually means "running a pool" in practice, because if you can produce shares, then nothing stops you from just using these shares to build your own chain.

When it comes to min-difficulty rule, only one thing needs to be changed: ASIC blocks should always reorg CPU blocks. It is a design mistake, that 20 minute rule forces people to mine min-difficulty blocks, instead of just allowing them to do so, and cleaning up their blocks automatically, when ASIC miner will come, and unstuck the chain.

[LoyceV]

What would it take for me to create my own testnet4 chain?

For testnet5, everything is ready, it is only a matter of launching it, and mining the Genesis Block. It is not yet deployed, because simply nobody did it yet.

That feels more complicated than running the existing Bitcoin Core testnet4 with my own fork.

When it comes to min-difficulty rule, only one thing needs to be changed: ASIC blocks should always reorg CPU blocks. It is a design mistake, that 20 minute rule forces people to mine min-difficulty blocks, instead of just allowing them to do so, and cleaning up their blocks automatically, when ASIC miner will come, and unstuck the chain.

That would be annoying if the ASIC block takes 3 days, and 50 full CPU blocks are wiped out, removing all confirmations for most of them.

[ABCbits]

It might be a good idea to just use the latest network.

Then use testnet5. As long as they didn't list it yet, it should be good enough: https://github.com/bitcoin/bitcoin/pull/35861

And if testing with multiple node or device is actually required, custom signet probably works good enough, https://en.bitcoin.it/wiki/Signet#Custom_Signet.

It would be great if Bitcoin was only used for regular transactions to avoid these things

If we would have the chain, where P2PK would be the only allowed address type, then it would be much harder to sync, and we would still have JPEGs.

Yeah, it require different TX design. FWIW, Grin TX design is probably works best to discourage adding arbitrary data on TX, see https://forum.grin.mw/t/public-transaction-data-is-a-huge-risk-vector/10426/14.

[athanred]

That feels more complicated than running the existing Bitcoin Core testnet4 with my own fork.

Quick lesson, how to deploy testnet5. First, mine the Genesis Block, and put its parameters in the code:

Code:

$ git diff HEAD~1
diff --git a/src/kernel/chainparams.cpp b/src/kernel/chainparams.cpp
index 8279fcf0a6..8712b06e2b 100644
--- a/src/kernel/chainparams.cpp
+++ b/src/kernel/chainparams.cpp
@@ -474,7 +474,7 @@ public:
         consensus.SegwitHeight = 1;
         consensus.MinBIP9WarningHeight = 0;
         // Testnet 5 has a raised minimum difficulty of ~1,000,000, i.e. nBits 0x1a0fffff.
-        consensus.powLimit = uint256{"0000000000000fffff0000000000000000000000000000000000000000000000"};
+        consensus.powLimit = uint256{"7fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff"};
         consensus.nPowTargetTimespan = 14 * 24 * 60 * 60; // two weeks
         consensus.nPowTargetSpacing = 10 * 60;
         // Testnet 5 has no minimum-difficulty exception.
@@ -516,17 +516,17 @@ public:
         // satisfy its own nBits and the node cannot run on testnet5 yet.
         const char* testnet5_genesis_msg = ""; // BIP95 leaves the message empty
         // TODO: use a recent mainnet block hash as the pubkey
-        const CScript testnet5_genesis_script = CScript() << "000000000000000000000000000000000000000000000000000000000000000000"_hex << OP_CHECKSIG;
+        const CScript testnet5_genesis_script = CScript() << "0200000000000000000001c8ed555ea374d64b1c2f2893a54ad1a50e07f8691487"_hex << OP_CHECKSIG;
         genesis = CreateGenesisBlock(testnet5_genesis_msg,
                 testnet5_genesis_script,
                 0, // TODO: nTime
                 0, // TODO: nNonce
-                0x1a0fffff,
+                0x207fffff,
                 1,
                 50 * COIN);
         consensus.hashGenesisBlock = genesis.GetHash();
-        assert(consensus.hashGenesisBlock == uint256{"e9a8fd2b0c78e01702e9ede3f79f2217142c25fb70e22dd86bff2879afd5761e"}); // TODO
-        assert(genesis.hashMerkleRoot == uint256{"b3f9a8c19683a9bb0475e37b13a77ed3051965c642da2aef093c7fe3eb08e3bc"}); // TODO
+        assert(consensus.hashGenesisBlock == uint256{"229e7383443170924172140bc3f47430988b7c2185253e8de37c7b3177132004"}); // TODO
+        assert(genesis.hashMerkleRoot == uint256{"8a86ddfbf5f85885a33c3e52830eb631749dd579ae63d81a7377a2aee4aff264"}); // TODO
 
         // No DNS or fixed seeds defined for Testnet 5 yet.
         vFixedSeeds.clear();

Of course, in practice you would need 0x1a0fffff difficulty, but if you just want regtest-like network, then you don't have to care about mining, and you can use CPU difficulty. It will adjust anyway. Then, recompile and launch it:

Code:

$ ./bitcoind --testnet5 --noconnect
...
2026-10-05T07:31:57Z UpdateTip: new best=229e7383443170924172140bc3f47430988b7c2185253e8de37c7b3177132004 height=0 version=0x00000001 log2_work=1.000000 tx=1 date='1970-01-01T00:00:00Z' progress=1.000000 cache=0.3MiB(0txo)
...

And then, you can see your Genesis Block from the CLI commands:

Code:

$ ./bitcoin-cli --testnet5 getblock 229e7383443170924172140bc3f47430988b7c2185253e8de37c7b3177132004
{
  "hash": "229e7383443170924172140bc3f47430988b7c2185253e8de37c7b3177132004",
  "confirmations": 1,
  "height": 0,
  "version": 1,
  "versionHex": "00000001",
  "merkleroot": "8a86ddfbf5f85885a33c3e52830eb631749dd579ae63d81a7377a2aee4aff264",
  "time": 0,
  "mediantime": 0,
  "nonce": 0,
  "bits": "207fffff",
  "target": "7fffff0000000000000000000000000000000000000000000000000000000000",
  "difficulty": 4.656542373906925e-10,
  "chainwork": "0000000000000000000000000000000000000000000000000000000000000002",
  "nTx": 1,
  "strippedsize": 184,
  "size": 184,
  "weight": 736,
  "coinbase_tx": {
    "version": 1,
    "locktime": 0,
    "sequence": 4294967295,
    "coinbase": "04ffff001d010400"
  },
  "tx": [
    "8a86ddfbf5f85885a33c3e52830eb631749dd579ae63d81a7377a2aee4aff264"
  ]
}

$ ./bitcoin-cli --testnet5 getblock 229e7383443170924172140bc3f47430988b7c2185253e8de37c7b3177132004 false
01000000000000000000000000000000000000000000000000000000000000000000000064f2afe4aea277731ad863ae79d59d7431b60e83523e3ca38558f8f5fbdd868a00000000ffff7f20000000000101000000010000000000000000000000000000000000000000000000000000000000000000ffffffff0804ffff001d010400ffffffff0100f2052a0100000023210200000000000000000001c8ed555ea374d64b1c2f2893a54ad1a50e07f8691487ac00000000

$ ./bitcoin-cli --testnet5 decoderawtransaction 01000000010000000000000000000000000000000000000000000000000000000000000000ffffffff0804ffff001d010400ffffffff0100f2052a0100000023210200000000000000000001c8ed555ea374d64b1c2f2893a54ad1a50e07f8691487ac00000000
{
  "txid": "8a86ddfbf5f85885a33c3e52830eb631749dd579ae63d81a7377a2aee4aff264",
  "hash": "8a86ddfbf5f85885a33c3e52830eb631749dd579ae63d81a7377a2aee4aff264",
  "version": 1,
  "size": 103,
  "vsize": 103,
  "weight": 412,
  "locktime": 0,
  "vin": [
    {
      "coinbase": "04ffff001d010400",
      "sequence": 4294967295
    }
  ],
  "vout": [
    {
      "value": 50.00000000,
      "n": 0,
      "scriptPubKey": {
        "asm": "0200000000000000000001c8ed555ea374d64b1c2f2893a54ad1a50e07f8691487 OP_CHECKSIG",
        "desc": "pk(0200000000000000000001c8ed555ea374d64b1c2f2893a54ad1a50e07f8691487)#2j4xsatp",
        "hex": "210200000000000000000001c8ed555ea374d64b1c2f2893a54ad1a50e07f8691487ac",
        "type": "pubkey"
      }
    }
  ]
}

Was it that hard? The most difficult part is probably just mining the block, but well, if you have any ASIC with a few TH/s, then you can do that, by just sending 80-byte block header into your device, getting the result, and then putting into the code. However, if you care about "just deploying the network", then it is easy. By editing "chainparams.cpp", and recompiling, anyone can launch testnet5 at any moment. If it wasn't launched yet, then maybe it is simply not yet needed.

[LoyceV]

Was it that hard? The most difficult part is probably just mining the block, but well, if you have any ASIC with a few TH/s

I don't have mining hardware, so I'll probably get a few MH/s. It will take days if not weeks to find a block.

By editing "chainparams.cpp", and recompiling, anyone can launch testnet5 at any moment. If it wasn't launched yet, then maybe it is simply not yet needed

If testing requires recompiling, I'd say it's more work than it should be for the average user.