Posts on X before Jan 17, 2026, 10:18:42 AM UTC
OSS is like a warzone rn
the normies figured out what a PR is and learned how to prompt claude code to send their turboslop through
every day i wake up i battle with which I believe more:
- openai has been destined to win since day 1
OR
- damn, everyone's catching up, it's so close
I feel the first when I am using a gpt 5.2 model, and the latter when I forget how superior gpt 5.2 models are (using opus)
Time to drive gpt 5.2 high instead of gpt 5.2 xhigh for the day
hearing it's much faster and in some cases "smarter" because it "doesn't overthink"
we'll see :)
hey gpt 5.2, can you review [feature done by claude]
"sure! let's see here, oh, this is... oh... oh my God... why would it... this part doesn't even... why are there four versions of this... did he just cram it all into main.go?... you need to report this slop to the police"
waste of time once ai controlled (already in testing/polished by EOY)hold - detect enemy aim/danger %, maneuver unpredictably towards target - detonate once proper maim/kill % chance threshold Not to mention, these advancements removes many pieces from the drone; lighter, faste… Less
waste of time once ai controlled (already in testing/polished by EOY)hold
- detect enemy aim/danger %, maneuver unpredictably towards target
- detonate once proper maim/kill % chance threshold
Not to mention, these advancements removes many pieces from the drone; lighter, faster, glitchlike movement, no need for impact sensors/rods, etc
Commenting on this because I find it weird the Chinese are training for this yet they know/are leaders in what I mentioned above
It's like training how to avoid arrows while you're also starting up production on mass producing guns
@GrandpaRoy2
A demonstration of anti-FPV defensive techniques in Chinese army training. The PLA has apparently been paying close attention to the use of FPVs in the War in Ukraine. x.com/GrandpaRoy2/st…
SOTA as of Jan 16 2026:
GPT 5.2 xhigh as orchestrator/planner/questioner
GPT 5.2 Codex xhigh as executor/programmer
Opus 4.5 if you like to talk and not feel lonely, but god forbid you let it touch your code/plan things for you
Everyone feels like software engineers are the only ones affected by what's currently going on in AI I'll tell you this much - I've been/am replacing/will replace soon, MANY more non-software-engineer jobs than software engineers Right now it's about 5:1 for me; for every job/c… Less
Everyone feels like software engineers are the only ones affected by what's currently going on in AI
I'll tell you this much - I've been/am replacing/will replace soon, MANY more non-software-engineer jobs than software engineers
Right now it's about 5:1 for me; for every job/contract I'd have given to a software engineer, I've also done the same replacement with 5 other non-software jobs/contracts
Will my engineered non-human replacements do as good/better work than their human counterparts?
no...
...t yet.
@bayeslord
people aren't seeing it. the role of business itself is changing. it's not just software
this is just common sense, and not understanding this immediately exposes lack of critical thinking if you think the solution to global warming (particularly the speed up caused by humans) is still "we need to consume xyz less", you need to update your reasoning regarding where … Less
this is just common sense, and not understanding this immediately exposes lack of critical thinking
if you think the solution to global warming (particularly the speed up caused by humans) is still "we need to consume xyz less", you need to update your reasoning regarding where we are headed in the next few years, let alone decades
plain question that should expose your thinking skills:
do you think our efforts are better spent towards reducing humanity's consumption (consumption of all senses; energy, material, service; pressure at all levels from gov-level to individual-level)
OR
do you think our efforts are better spent towards increasing humanity's consumption especially regarding resources necessary to not only mitigate but "solve" the core issues and worries of global warming (which would be a drop in the bucket of so many other solveable problems thanks to the "superintelligence" / super-systems we're already seeing the first sparks of today(!)
choice number two isn't wishful thinking, nor improbable, nor chance. though I can see how it may seem so if you haven't been paying attention to the cutting edge that cuts ever further, ever faster, every single day
@Plinz
I am not a "tech leader", and I am not speaking for any. I believe that global warming is a serious concern, but the ship has sailed 30 years ago. Current efforts to reduce emissions are no longer going to make much of a dent, and we must focus on mitigating effects now. I have l… Less
I am not a "tech leader", and I am not speaking for any. I believe that global warming is a serious concern, but the ship has sailed 30 years ago. Current efforts to reduce emissions are no longer going to make much of a dent, and we must focus on mitigating effects now. I have lost a lot of sleep about the topic when I was young. I understand the feeling that we must do something, urgently, but it is also important to psychological come to terms with the bitter reality that international coordination failed when it was still possible. (Btw, I lost a lot of sleep over this in the late 80ies and 90ies, when the public and media mostly did not care at all.)
if you're not mainlining gpt 5.2/5.2 codex, you're wasting your time, even if you think you're being productive with Opus 4.5. after many weeks of trying 5.2, 5.2 Codex, Opus 4.5, Gemini 3.0, I can safely say 5.2 Codex is the best programmer, 5.2 is the best planner. Opus 4.5 i… Less
if you're not mainlining gpt 5.2/5.2 codex, you're wasting your time, even if you think you're being productive with Opus 4.5.
after many weeks of trying 5.2, 5.2 Codex, Opus 4.5, Gemini 3.0, I can safely say 5.2 Codex is the best programmer, 5.2 is the best planner.
Opus 4.5 is a quicker but dumber programmer/planner.
Gemini 3.0 is literally unusable, absolute trash. In comparison, though. I guess if you're making a quick change here or there it's OK
I currently have 2 OpenAI Pro subscriptions ($400/m) and 1 Anthropic Max sbscription ($200/m).
Once my Anthropic Max subscription ends, I think I will just get the third OpenAI Pro subscription instead.
Hey @OpenAI can you make a $500/m subscription instead please, so we can all stop this account juggling nonsense?
gemini 3.0 pro is either super fast and works very well
or super fast and causes increased retardation when reading its reasoning as to why it made the worst decisions possible
not a serious model at all. slop
I think he misunderstood what "bone smashing" meant when he was reading about it on the looksmaxxing forums
@CatchUpFeed
A looksmaxxing zoomer entered a grocery store looking to pick a fight and made the life-changing decision to sucker punch a Mexican landscaper. It was a bad idea.
I'm dead center, this year I plan to finish atleast 2 steps higher :)
@marclou



