dont refuse to believe things,
things are real
nomadshiba⚡⚡
me@nomadshiba.me
npub1gkp4...ppqk
its me, you know
bip352 / silent payments: sp1qqf95gnrh453sncve4yal29zedxll3k48a0kcxhted66nnu5frcshwqmduw4x6a9ggvcmesah9qu32jx95ust635ng6xrd570mpu9hhmn7cpwpee5
My Apps:
- https://npub.name
- https://hashrate.place
- https://persona.nstr.app
- https://zapme.coffee
Soon:
- https://bitcoinatlas.app
- https://nstr.app
nostr is amazing, lots of different protocols so cool.
best part, you only download what your client cares about.
nobody has to store everything. truly amazing.
It's About to be Even BETTER 👀
A new chapter


can somebody help me pay the bills this month? im kinda stuck. anything helps.
lnbc4500u1p4gwqrgdqqnp4qg3swhjx35csgqnh79uktlxsq2qp3r2659ztqpvcv5nu7lsw0amucpp5uzgrmf5c68ln4f5484505p3lq7tujecd3lzat5gelr4wn7fsf3pqsp5mwtfzawc4fp8hyv4wezx6kj7tg52spc0843fcv3078ua99h3ekaq9qyysgqcqzp2xqyz5vqrzjqw9fu4j39mycmg440ztkraa03u5qhtuc5zfgydsv6ml38qd4azymlapyqqqqqqq62gqqqqlgqqqq86qqjqpt5rmn0fe5c96gvt9yfp88e36w8t98s0pduryeyv3dqej24uv8m4h6m5n2varxryhufvf8e08ggallewj9nvx50mh9jth70x7nqdnqspl3jgk8
vengeance
GM
GN
i don't use GPT models anymore if i do i do it from free plan, i dont wanna give OpenAI demons any money.
sometimes i use Claude because work pays for it.
but mainly i use GLM5.2 for coding so here is my experience.
Claude Opus 4.8 close to or better than GLM5.2 in terms of understanding things, and following guidelines, reasoning about exiting code and existing code style etc. its better at debugging complex bugs as well. if you are not trying to one-shot find a bug cause, GLM5.2 is great at many tasks, tho sometimes it gets into these retarded personas where it doesnt understand anything, doesnt do exactly what you said, acting like a total idiot. but most of the time its smart, understands the codebase well, does everything correctly. etc. also has a big context so that also helps a lot.
Claude Sonnet 5 is 100% worse than GLM5.2 for coding, it makes the dumbest decisions, its basically bad, even for simple refactors, which is surprising. dont use it. maybe use it while brainstorming or talking about things by yourself. for coding it cant do shit, doesnt understand stuff. doesnt understand code. you keep have explain stuff. it gets wrong ideas from the code.
cool thing about GLM5.2 it can just go on and on and on doing something without even feeling 10% of its context. since it does some mistakes, its important that it has bigger context because requires some more corrections.
didnt try new deepseek models tho, maybe they perform better. they say its clause to fable, so idk.

one of my goals is a UAHF
i kinda got used to it but sometimes my internet connection is too slow, i struggle to like notes. or post/reply stuff.
annoying sometimes.
wish people who code nostr clients cared more about bandwidth and data.
bitcoin is not in some far away place, in the cloud, or some site online.
bitcoin is within you. bitcoin runs on your computer.
you have a record of how much money/energy every other user of bitcoin has, locally, on your computer. and you grant sats to everyone who deserves it with in your eyes, and reward them.
bitcoin is decentralized, not distributed.
everyone has their own instance of bitcoin with its own records.
