You speaking with the guy that has used 12 different IDE's and ended-up in neovim so I see you. I've had ClaudeCode, OpenCode, Codex, Orca, Conductor, Herdr, VibeKanban, you name it G, I'm now using Pi with my own lil setup with 3/4 extensions
It's been 3 days now LMAO but I've reached a point where it's sort of hands-off and properly delegates and returns to the orchestartor
Looking to optimizing is veeryyy tempting, but I'm forcing myself to be very intentional with my AI usage
Not trying DeepSeek v4 Flash:
@fr0styJS@jayair crazy innit, it's incredible value for money.
I was thinking of going to a bigger 100β¬ model but now just sticking with GTP 5.6 Sol for planning and DeepSeek for working
I'm Portuguese, and when I started working 10 years ago I earned 450 euros a month as an iOS Developer as an 18 year old high-school guy.
Tho I've increased my salary by a good margin, it's still weird in my brain to spend 200β¬ a month on a subscription
Great news, now my 5% usage in 5-hours will be 2.5%
DeepSeek V4 Flash is a game changer in terms of Intelligence + Speed.
For me, way more impressive than GPT Sol 5.6
I understand those who can dump money in marginal gains, but DeepSeek V4 Flash can be used worldwide, no longer only for rich countries
@heyitzami Quality of Life improvements I would say, things that would improve my sleep, exercise, health etc
I donβt even want anything tbf, just do shit for the love of the game
Maybe welding or electrician lessons lmao
Reporting back: it's so good, I'm keeping it
One of the things that used to annoy me was all this fake gaslighting language ("this is the smoking gun")
Or verbose, waffling technical prose that is hard to parse
And now it's gone π§ββοΈ
No matter what you do, never stop being yourself because others find you insufferable, fuck them, they can mute/ignore you, specially in social media
Do your thing bigman
@DanielLockyer 100%! based on current costs of modes I thought it was a subsidy to bring people to deepseek/china models, but the fact they were able to host it without losses is very interesting, not sure about scaling it like deepseek did, but I guess we will see very soon