I actually use openrouter to access many models, and we can have web search also enabled with it: https://openrouter.ai/announcements/introducing-web-search-via-the-api

Was there anything besides this feature which makes you prefer Requesty over open router?

I had similar experiences with ChatGPT 3.5 in the past. I was feeding it matrices and it was consistently generating wrong answers with calculations. I swear I would never use it again. Then came ChatGPT 4.o. I think for people who want to learn a foreign language it became useful and funny. Plus it became useful for math, at least for me. I do calculations and I check my results with those of ChatGPT. If it matches, I consider it is likely correct, else I re-check. After re-checking if it still doesn't match, I keep my calculations and ignore those of ChatGPT. This way I never got disappointed anymore with it, since I use it as a tool to do double checks.

For coding, I feel like Claude can also give valuable answers sometimes better than ChatGPT. Recently I asked both to build an algorithm to convert a Golang algorithm to Python to encrypt passwords... And the results were consistently 100% totally wrong. Very disappointing isn't it? But in this case I was double checking with a good Golang version by decrypting the encrypted password. So with this kind of check I could immediately spot incorrect code. Disappointing but fine since the only negative impact was to lose a little bit of my time working on it. Bottom line regarding code, Claude can be good as well but currently more expensive, and regardless this is really important to have unit tests or some kind of test to double check results.

So for coding I use it more like a very advanced auto-completion tool. E.g.: I want to read a CSV file and parse it a particular way, then it gives me the code right away. Or I want to write a SQL statement to show some statistics about a metric, it writes it. Its use is tremendously good to save time but everything has to be constantly double-checked by testing the code, or by doing a quick visual check, or by comparing carefully results if money is involved and it is extremely important that the code must be right.

For me the main issue with Zeus is the battery discharge rate. I think it needs improvement on this side. For example with Tailscale I can access my Umbrel node from mobile and save battery time without using Zeus. The idea of running a lightning node on mobile is really appealing though. My battery is an relatively old 4000mAh.

This could explain why I see so much Euro trades on Robosats. It has been more than 2 years I think I have left CEXes. Now I have peace of mind and don't have to think about the next time I will have to renew my KYC info or to think about what to do when my account won't be working anymore. In the past I was a Swan Bitcoin customer, I think the platform had 3 times issues in the span of 1-2 years because of some new regulation or partner/internal issues.

Actually I didn't read this post but this one and it was more detailed: https://gnusha.org/pi/bitcoindev/9c50244f-0ca0-40a5-8b76-01ba0d67ec1bn@googlegroups.com/#t

I read this post yesterday, together with the one from Peter Wuille on the mailing list it was the most interesting reply I read. I didn't know there were miners who were offering to directly broadcast their non standard transactions through their web interface: https://slipstream.mara.com/

Also his findings show that spam filtering in the end is detrimental to a healthy P2P network. This was something I was not aware of as well, I thought filtering would genuinely impede transactions from getting into the network. I didn't know also about Libre relay also which voluntarily takes the counter side of Bitcoin Knots.

To me this is a strong argument against spam filtering. Ultimately it looks like we will have to deal with non-payment data on the network. From a utilitarian perspective I still struggle to see why using Bitcoin would make sense for other use cases than money though.

I am trying to understand the joke, what do you mean?

Just recently I used it on Jam.

Friendly collaborative cancellation is still a positive outcome, so 98.4% for me. Sometimes platforms to do fiat payments just don't work as expected and someone may need to cancel. I had a trading peer who had to cancel once. I bought bitcoins from someone else. So positive outcome.

I see. If the data on-chain would not increase, then we could expect that the number of nodes would not be negatively impacted. I would be more convinced by those in favor of the change then.

Ultimately regarding spam, I saw a post from Sjors I think saying that to get rid of useless data (by useless I mean unrelated to Bitcoin as a payment network), a deeper change in the consensus mechanism or soft fork would be needed anyway.

The most convincing argument against the change I have seen is the one for node runners. We could forecast an increase in data consumption on the blockchain sent. So this would force us to store data which could be stored elsewhere and accessed in my view more efficiently. And so this would end up in a smaller group of node runners. I don't know if this argument is correct but to me it is the most compelling one.

On the other side, the argument of Pieter Wuille about network fragmentation (his argument being about miners bypassing the public network) looks to me compelling. It was the 12th post here: https://groups.google.com/g/bitcoindev/c/d6ZO7gXGYbQ

I don't know what the best thing to do is, as I am not knowledgeable enough about the implementation of Bitcoin, but I hope knowledgeable people keep debating without using emotional arguments and keep neutrality. To me what differentiates Bitcoin from other payment networks is its neutrality. You can be a North Korean hacker who stole money or a poor guy fighting inflation, both have equal opportunity to make transactions accepted on Bitcoin. In this regard, using the spam argument seems to me a less important one as it uses the nature of the transactions (and in turns lacks neutrality).

I understand the frustration and don't want to justify slow response times. But I feel somewhat targeted as I would estimate it happened 2/15 trades this month that I accepted an offer and couldn't reply right after due to personal reasons. 90% of the time I reply right away though. I would suggest waiting for at least 1h before doing a collaborative cancel and considering it is a scam. I never have used Amazon gift cards though and don't want to undermine your post, it is good to know that people may be trying to scam with it!

Sorry, misread the question. I will delete my answer.

I highly agree with this comment. From a philosophical perspective it reminds me the discussion 10-15 years ago about ACTA: https://stallman.org/articles/acta-freedom.html

I remember 3-4 years ago I saw a report from a Bitcoin company stating the proportion of bad actors doing transactions is not significant. Unfortunately I can't remember the source of the document, so instead let me quote ChatGPT: "According to a 2021 report by Chainalysis, illicit activities accounted for only 0.15% of all cryptocurrency transactions that year, totaling approximately $14 billion". 0.15% is not significant. Condemning everyone for terrorism by imposing some form of bureaucracy doesn't work.

We should worry more about fraudulent activities inside government institutions (like... USAID) which impact everyone. Today politicians may want KYC for terrorism, yesterday the trend was spying on everyone for children. In El Salvador it seems they didn't need KYC or another form of bureaucracy to arrest criminals.

I expected corruption with money from USAID in Latin American countries after it was sent. I am surprised to see that it was used for a totally different purpose even at its origin.

I see, thanks

Did someone successfully installed it on Nixos? The version there is quite old.

很好的

500 sats sent. How is China so far? Is what we hear about surveillance is true?