Hacker Newsnew | past | comments | ask | show | jobs | submit | tiahura's commentslogin

I'd like to see a nethackbench.

Do people really use claude code or any other agent with a paid api key? Why? Why wouldn't you just get Claude Max?

I wouldn't use Claude API Key pricing, but I also wouldn't get a Claude Max sub unless it was the only AI tool I used.

Antigravity / Google AI Pro is much better value, been using it as my primary IDE assistant for a couple months and have yet to hit a quota limit on my $16/mo sub (annual pricing) which also includes a tonne of other AI perks inc. Nano Banana, TTS, NotebookLM, storage, etc.

No need to use Anthropic's premium models for tool calling when Gemini/MiniMax are better value models that still perform well.

I still have a Claude Pro plan, but I use it much less than Antigravity and thanks to Anthropic axing their sub usage, I no longer use it outside of CC.


Counterpoint: on the $20 monthly account I would hit my 5 hour limits within an hour on antigravity. I end up spending half my time managing my context and keeping conversations short.

Rate limits mostly - plus claude code is a relatively recent thing but sonnet api has been around for a while with 3rd party apps (like cline). In those scenarios, it was only api.

It will at least identify the key disputed items and claims. Chatgpt will routinely balk on topics from politics to reverse engineering.

Even more strange is that sometimes ChatGPT has a behavior where I'll ask it a question, it'll give me an answer which isn't censored, but then delete my question.

Claude code, nethack, and tmux are fun to experiment with.

To be fair, there were a few folks on x avh and the other typical sites who guessed the cause pretty quickly.

Like most things, guessing and proving require vastly different efforts. In aviation, a few more orders of magnitude than most.

If a few thousand people speculate some of them will get it right. It means nothing.

The announcement says existing connectors work, but only Claude for chrome does.

Lawyer here. AI has taken over my workflow.


Good thing nobody listened to the LeCuns saying cars were a deadend back in the 1910s.


There is some clinical evidence of other strain effect. See pediatric dermatology article from a year or two ago.


How does Tesla FSD respond to inactive traffic control lights?


Coincidentally we were on the Robotaxi during the black out (didn’t know about it, we were going to Japan town from the Mission). Noticed that it navigated through the non-working traffic lights fine, treated it like a stop sign junction. One advantage of building unsupervised system from public version that had to deal with these edge cases all around the country.

Though the safety driver disengaged twice to let emergency vehicles pass safely.



https://x.com/edgecase411/status/2002630953844552094

Looks like it treats it as a 4-way stop. Is this because Tesla has more training data?


I'd default to assuming it's the respective roadmaps for Waymo and Tesla differed on which things to implement when, not training data, that results in the two behaving different.


50/50 bet it would either go right through or treat it as a stop.

Don’t think I have had a totally inactive light. I have had the power is out but emergency battery turned to blinking red light, and it correctly treats as a stop sign.


> Is this because Tesla has more training data?

Its human takes over. FSD is still Level 3.

(Robotaxi, Tesla's Level 4 product, is still in beta. Based on reports, its humans had to intervene.)


FSD is level 2. Level 3 doesn't require the human driver to monitor the outside environment, only take over when requested. Tesla also doesn't report data from FSD under L3 reporting requirements anywhere in the US.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: