Which Is the Best Self-Custody Lightning Wallet?

I’m conducting my second test of self-custodial Lightning wallets in Zimbabwe, to see which might be best for any rural area. Given that none of the available self-custodial wallets are developed by Africans, they are rarely used or tested here. On top of that, Lightning is still new and complex, building mobile apps that are running a node of the network on your phone is challenging.

I am here to find out what works and what does not. As an educator, I recommend tools for everyday users, who want to send small payments. These tools have to be reliable, otherwise people might conclude that Bitcoin isn’t for them.

I opened my first Lightning channel on an app called BLW in January 2019. I’m used to testing new tools, that’s why I have no problem if challenges and bugs arise. But regular users may not be as forgiving, that’s why I aim to only recommend tools I’ve personally tested.

In 2023, the only two self-custody Lightning wallets available were Breez and Phoenix.

This year I had more choices and tested: Blixt, Mutiny, Green, Zeus and Phoenix. I also included the Wallet of Satoshi, a custodial wallet, just to see how it performs in comparison. I was familiar with Phoenix and used Zeus connected to my Voltage node earlier. Mutiny, Blixt and the Lightning integration in Blockstream’s Green wallet just popped up in the last few months, so using them was new for me too.

All of the wallets have been undergoing huge changes and will continue to evolve. I had to download Mutiny manually and install a special .apk file used on Android phones. For the Green wallet, I asked BlockStream to unlock its experimental Greenlight feature in order to test it.

My goal was to find out which of the wallets are the best in sending and receiving bitcoin in a reliable, fast, easy to use manner and to compare the costs involved.

Before heading to the rural areas (I intended to test the wallets under suboptimal conditions), I set up the wallets and opened a Lightning payments channel in Harare as I anticipated low internet signal challenges. Based on my year-long experience in Zimbabwe, I’ve learned that even with decent internet speed, the upload and download of files as small as 2 MB can be problematic.

I used the internet at a friend’s house to use the fastest internet provider in Harare called Liquid, which is also the most expensive with a monthly cost of $300 for unlimited usage. My speed test results for download were 92.7 Mbps and upload 14.6 Mbps. Even with that fast internet it took me some time to install the apps, secure the private keys and open a Lightning channel to be a part of the network.

To ensure consistency, I followed a test protocol for each wallet. On Dec. 26 and 27, I installed the wallets and sent 100,000 sats (~$42 at current prices) to open a Lightning channel. A channel on the Lightning Network needs two transactions on the Bitcoin blockchain to be set up. That way the channel is constantly connected to the blockchain, which is why the bitcoin on the Bitcoin blockchain and bitcoin on the Lightning network are the same. There is no difference in the value, it’s the same unit, not another token or asset.

There are several ways to open a channel: sending on-chain bitcoin, transferring Lightning directly from another wallet or through a swap from Liquid or some wallets are offering buying inbound liquidity. I tried them all to send the first bitcoin to the wallets, affecting set-up speed and fees I had to pay.

I used a Lightning payment to open a channel on all wallets, except for Blixt, as I only found the on-chain option there (except for liquidity providers). Seemingly, Blixt is opening the channel with a second on-chain transaction while the other wallets conceal that second transaction somehow, because the channel opened immediately after the first transaction confirmation. Not with Blixt: I sent the first transaction and the confirmation took some hours, it was evening already, so I had to leave the friend’s house. As soon as it was confirmed, Blixt carried out the second on-chain transaction, which took again several hours because the mempool was full. When the second transaction was confirmed I was back in the house where I’m living, with internet speed only around 3 Mbps download, 0.26 upload and a Ping over 300ms. Blixt wasn’t able to open the channel in these circumstances. I needed to go back to my friend’s house to finish the channel opening. It’s hard to say how other wallets would have performed under these conditions.

One BTC is pegged to one L-BTC. This means if you peg-in 0.1 BTC to Liquid, the resulting 0.1 L-BTC are of the same value as BTC. Compared to self-custodial Lightning there is more trust involved in using Liquid, because you have to go through intermediaries for the peg-in and peg-out and trust a federation of 15 companies, who are the signers for Liquid transactions, whereas on Bitcoin and Lightning you don‘t have to trust any intermediaries. The upside of the Liquid blockchain is that the transaction speed, privacy are higher and currently the fees are lower than on the Bitcoin blockchain.

I compared all network fees, transaction, and service fees that applied with the different methods of opening a channel.

The cost of the incoming transaction to open the channel was the highest with Blixt as I used bitcoin on-chain at a time when the average fee rate was 110 sat/vByte ($0.000054). Channel opening was the cheapest when using Lightning directly from another LN wallet. The route using Liquid and Boltz was a little bit more expensive than using Lightning.

The available balance after opening the channel varied significantly. Zeus had the lowest balance of 52,500 sats, while Green had the highest at 97,500 sats. This difference was surprising, especially since I used Liquid for both. Green was the clear winner with a cost of 3.5%, whereas Zeus was by far the most expensive wallet with 48.5% of the funds spent in opening the channel.

Mutiny offered the biggest channel capacity and the highest receivable amount with 100,000 sats, while at the same time it was the second most expensive. However, a drawback of Green was its limited incoming capacity of only 4,133 sats.

After the initial set-up, I set out to do the payment test on Jan. 1, 2024. I strolled through the scorching heat, making my way up to the kopje, navigating through rocks, grass, bushes and trees 30 km out of the capitaI Harare. My friends were captivated by the beautiful flowers that bloom as soon as the rainy season begins. My focus, however, was on finding a decent internet signal for my two phones. Having prepared the test back in Harare, I was eager to carry it out in this setting.

After 30 minutes of searching, I realized that the network coverage was inconsistent, regardless of whether I used NetOne or Econet. I was able to send WhatsApp messages, but as soon as I wanted to download an image with 0.5 MB, I had to wait or turn airplane mode on and off to be back in the game. Finally, I settled on a shaded spot with a stable NetOne signal to set up a hotspot for my other devices. I prepared my devices – my Google Pixel 4, iPhone 13 Pro and my iPad Air (3rd generation). The iPad was included because I already had Phoenix wallets on my other two devices, and for this test, I wanted to start with a fresh installation.

Now seated on some rocks, surrounded by trees and bushes, as I fend off large ants attempting to crawl up under my trousers I found 4.21 Mbps download and 0.36 Mbps upload. Glad that I set up the apps already a few days earlier.

My speed test results showed a ping of 90 and a download speed of 4.6 Mbps and an upload speed 0.5 Mbps. When I started testing the wallets, the Phoenix wallet opened immediately. The Green wallet took 13 seconds to open and the Mutiny wallet needed 18 seconds. However, Zeus didn’t open at all, even after a 150-second wait. Similarly, Blixt opened but failed to sync the channels and wallet, so I couldn’t use it. As expected, the custodial Wallet of Satoshi opened immediately.

My first test was to send 50,000 sats from the different wallets to my Phoenix wallet on my iPhone. So, the Phoenix wallet on the iPhone was on the receiving end. Unfortunately, I could not send from Blixt because, as I said before, it did not sync at all. Zeus was also unusable since it wouldn’t even start up. When I attempted a payment with Zeus, it failed after a five minute wait. The Green wallet’s payment attempt was unsuccessful too, displaying an error message after I waited for two and a half minutes. With the Mutiny wallet, I almost wanted to stop but after 50 seconds, I was able to send the payment successfully. The Phoenix wallet was the quickest, completing the transfer in just three seconds. And interestingly, the custodial Wallet of Satoshi was slower than Phoenix, taking nine seconds for the payment to go through. I didn’t anticipate this result, but I found it amusing.

The next test focused on receiving 30,000 sats from the Phoenix wallet on my iPhone, which had previously served as the receiving wallet. The results were somewhat similar to the previous test. The time until the payment was sent from Phoenix was three seconds and it showed up in Mutiny, Phoenix and the Wallet of Satoshi immediately. Blixt did not work. With Green, the payment failed after I waited for 35 seconds. Similarly, Zeus also failed to receive the payment after I waited for one and half minutes.

For this test, I sent 20,000 sats to a Lightning address, using the Wallet of Satoshi as the recipient since I had access to its Lightning address. The results were somewhat consistent with previous…