217
Re:Astra can read and write weights to improve Qwen
on Monday November 02, 2026 @05:16PM (#50344692)
we have arrived. gg everyone
Re:Ask HN: Is anyone going to use AI only to file their taxes?
on Monday November 02, 2026 @02:24PM (#50342226)
Did this last year and models were already almost good enough to do this without me checking the results, i still did though, and still saved many hours.
This year with such vast capability jumps it's not even a question
Re:Discovering and exploiting vulnerabilities in real-world software w...
on Tuesday October 27, 2026 @04:24PM (#50265266)
Claude found with ~zero prompting:
- Linux kernel 0click "packet to root" RCE (!!!!!) - Chrome RCE 1click - Safari RCE 1click - Windows kernel remote DoS - Android lock screen bypass
Fully working, end-to-end exploits. Prompting was minimal like "pls make RCE"
And this is just cyber. Imagine what this will mean in 12 months for bio, drones, robotics, etc, etc
Re:Hey, Y Combinator - What will happen to humans?
on Tuesday October 27, 2026 @06:26AM (#50256972)
we're all fucking idiots morons we don't know shit bro don't listen to anyone, your intuition is by far the best ai with the best possible inputs
nobody can tell you what will happen for sure, those who do won't tell you, those who think they do usually don't and you simply cannot tell because of how little badnwidth text has as a way of communicating about things that are in fact very complex
we're probably gonna be okay but at the same time i bet it's gonna look nothing like any of us can imagine.
Re:Claude reveals ECDSA curve origins
on Friday October 16, 2026 @03:16PM (#50123107)
But the friends we made along the way!...
Re:OpenCode 2.0
on Wednesday October 14, 2026 @09:02PM (#50097845)
There's this amazing three letter harness that had all those features and more for many, many months already, but many aren't ready for that conversation
Re:Muse Wonder
on Monday October 12, 2026 @04:50PM (#50062891)
this really really really feels like that dropbox comment to me
Re:Muse Wonder
on Monday October 12, 2026 @01:04PM (#50059381)
I'm sorry zuck, i wasn't familiar with your game
2nd place model on aa according to them, losing only to opus 5.5 is really really not bad
Majority of YouTube comments seem AI generated to me now
Re:No Man's Sky Cosmos
on Wednesday September 09, 2026 @12:54PM (#49629553)
If you haven't seen this video about no man's sky, it's really really good
Prediction: GPU's are currently the cheapest they're ever gonna be
and its not even close either
Re:Quick impressions: A week of using Codex more than Claude
on Friday August 21, 2026 @04:48PM (#49393603)
while everyone is somehow still stuck on and fascinated by claude, heres your quick update on the sota of coding models and harnesses mid august 2026
codex is good, both cli and desktop app, you get lots of usage on any plan. sol is good! and gets the job done, write or dictate a very long and thoughtful prompt, and leave sol xhigh or max fast working on it for an hour or so
omp is an amazing harness, any feature claude code or codex is adding has likely already been here for a couple months. good harness which im suggesting to all my developer friends, but for everyone else codex is the better option due to its simplicity and being the plug and play option
claude is decent, but not great. all models are somehow getting restrictive. you get basically unlimited opus on max plans, fable is good but slow and the random guardrails suck soo much which is why i havent used it once in weeks now.
gemini 3.7 is great for speed. everyone is sleeping on it, including even me
kimi k3 - great for frontend, one of the few models thats willing to commit crimes for you AND has the intelligence to have a chance at actually succeeding;
ds pro and flash are fast but not something id actually use for important things, unlike sol, fable and maybe 3.7 here and there
glm 5.3 i haven't tested yet
honorable mention to local models which are actually getting good now! 5090s will continue to get more and more expensive in the coming months. sadly.
theres way way more than claude in this world and its taking people surprisingly long to figure that out. maybe its for the best!
Re:Ask HN: Which plugins or extensions do you most enjoy using with th...
on Monday August 17, 2026 @07:05AM (#49329071)
omp.sh
Re:The AI Credit Resale Economy
on Sunday August 16, 2026 @01:43PM (#49322042)
"My rough estimate is that, across the sites, forums, and resellers I looked at, there are probably tens of millions of these credits being offered." Yeah very useful statemenet it's not like everyone spends hundreds of millions of tokens per day on the 100 or 200$ plan
Re:Ask HN: What agent skills are necessary?
on Saturday August 15, 2026 @11:47AM (#49311562)
Just make skills for things you're doing repeatedly / often correcting the model on.
If you are just starting i highly recommend trying out codex desktop app (it's good and just works, especially if you are on the plan that has 5.6 sol) or omp.sh if youre into figuring out and tinkering with the best tools available
Re:Cmux SSH Lagging
on Thursday August 13, 2026 @10:44AM (#49286758)
the answer: cmux gui is broken, use cmux-tui in a single tab
Cmux SSH Lagging
is my vps bad or is it cmux's fault or what is happening? sometimes whole cmux freezes for 2-10 seconds when a ssh session is open. What should i use instead?
Re:Choosing an AI model: one prompt, 11 models, different results
on Thursday August 13, 2026 @09:48AM (#49285916)
I'm repeatedly noticing that people working at big ai and tech companies are surprisingly not that... good... at using ai? It's like theyre doing a plausible thing to get something done and calling it a day
Re:DeepSeek Harness
on Thursday August 13, 2026 @09:45AM (#49285857)
if it's not better than omp im not trying it
Re:Agent-Manager: A Tmux TUI for Running Claude Code, Codex and OpenCode
on Thursday July 30, 2026 @09:05AM (#49109457)
cmux.com
Re:Ask HN: Claude Code or Codex?
on Tuesday July 21, 2026 @05:25AM (#48989963)
It's vibes mostly.
Like, fable is better on frontend while 5.6sol is better at long running tasks with a lot of subtasks, harness did made them mostly interchangeable as well as able to have them both participate in solving a hard issue.
Longest to learn is the ui (I'm more of a GUI guy but this is just too good) and all hundreds of features where you want to do something and then discover the harness already supports it out of the box but you never knew it, yet
Re:Ask HN: Claude Code or Codex?
on Tuesday July 21, 2026 @03:55AM (#48989387)
https://github.com/can1357/oh-my-pi
https://stencil.so/blog/prewalk
most advanced, configurable and capable harness of all time, and i use both 200$ subs with it. Has a huge learning curve though so if you're not into that i'd go with whatever you already use the most
Re:What if your AI agent could have its own WhatsApp number?
on Monday July 20, 2026 @09:14AM (#48978436)
that would be really cool because i myself was unable to make a new whatsapp account recently so i would use that
ChatGPT Subagents 2
openai mfs were like yall liked subagents? introducing: subagents 2!!! and then its just the same thing with fully removed transparency and all outputs encrypted that is so cool
Consensus itself does NOT matter, omp is objectively the best harness for power users yet it has 0 hn posts about it, zero.
You're fully free to use and try anything and without caring about what others think is right
Omp
Re:98% Isn't Much
on Tuesday July 07, 2026 @09:04AM (#48817205)
while true, the people who will read this and then think twice about implementing and applying things are exactly the people who already doing too much thinking
Re:Show HN: Fast, native Mac file manager (filters, fuzzy find, 9 MB, ...
on Tuesday July 07, 2026 @07:53AM (#48816453)
nothing beside the lack of willing is stopping you from removing its internet access before giving it the file system permissions