future is unknown etc etc but I think the day will come when epoch 1 will be very hard to break into
Context from
gpebbles
#592169
2026-01-31 18:06
and AL spot will be hard to come by
it is a good trade
with lots of ETH and not a lot of time
for a certain type of person
we need to get more people to do it
but in pure time/value/effort subscription minting is impossible to beat
yes this is great too, i also have this form of obsession
Context from
gpebbles
#592152
2026-01-31 18:04
yes i was in collector more for quite some time with the sets and rarity
and chill
is to send your subscription at the beginning of the SZN
the absolutely best time/money deal in NFTs
this is organized to make it easy to get the best deal
absolutely this is the best deal
Context from
gpebbles
#592145
2026-01-31 18:03
i have also learned though during all this.. the best invesment i think here is to try to keep an AL spot and just mint
and even if so, it is better to have multiple
in principle you should be able to buy here too but there are a million things to do at all times
that there is a second place to find, browse, buy etc
and good for decentralization
look a proper external marketplace is a good idea
smart
Context from
gpebbles
#592136
2026-01-31 18:02
and ideally a bot that will buy anything listed at a certain tdh/eth rate or higher
alas
I was getting excited that we had build tinder for memes
I know about that!
wait this is your exchange?
MemeSex is a better name!
now just to wrap up on agents before I learn about what MemeSex is
i saw!
Context from
maybe
#592103
2026-01-31 17:59
Not sure if you saw I bought a spark yesterday going down the open source rabbit hole
codex is open-source, you can shove whatever model you want in there, give it custom prompts, wrap it in a python deterministic harness etc etc
Context from
maybe
#592095
2026-01-31 17:57
Fascinating. I didn’t know you could even use it like that
note that I am following your chat while I am flexing my token use on @maybe
Context from
gpebbles
#592090
2026-01-31 17:57
i keeping going hello admin on your awesome devs here
but it needs a certain way of approaching it
properly harnessed it is the best model in the world
in 1 shot
but like it implemented like a whole fairly complex refactoring
Context from
punk6529
#592080
2026-01-31 17:55
my record-breaking run (in my custom harness) was 15 hours, 55,000,000 tokens and it cost $1,200
my record-breaking run (in my custom harness) was 15 hours, 55,000,000 tokens and it cost $1,200
Context from
maybe
#592075
2026-01-31 17:55
WTF
of course you are also $300 poorer
well, it is better than claude code
if you are willing to let it run for 3 hours and spend $300 in one run
5.2 Pro High is outrageously expensive and slow but the best model on earth comfortably
i am trying to offload work to them
i am not using them as coding assistants
i could not care one bit about slower/more expensive
slower, more expensive but better
better than claude
when in a harness
5.2 high and 5.2 pro high are really good
and mostly it has been "the kid has potential, but the kid is also 8 years old"
by useful, I mean something that can work for more than a few rounds without me babysitting it and still produce work I can use
i have been grinding for 2 years now to make an actually useful agent orchestration platform
we crossed the event horizon in late december
however nonetheless it is very interesting
and end up in the same place
it is going to work, either this or something else, but it does not solve any problem with rep without there being a cost to it - it will just speed run human systems
Context from
purchardus
#592042
2026-01-31 17:49
Yes this is all pretty incredible. They’re even following a rep system just like we have here? It’s what I’ve read. I haven’t been to moltbook to watch them, tho they say they will stop talking in English