News for hackers. Stuff that matters.

vhayda

Re:How many of the 170k English words do you know? on Friday June 19, 2026 @05:23PM (#48603400)
The longest answer choice is correct 80%+ of the time, when it should be closer to 25%. I was able to breeze through unfamiliar words just by picking the longest option every time…
Nice! It needs some refinement and a macOS version.
Re:ChatGPT Pro on Thursday December 05, 2024 @02:03PM (#42331364)
Yesterday, I spent 4.5hrs crafting a very complex Google Sheets formula—think Lambda, Map, Let, etc., for 82 lines. If I knew it would take that long, I would have just done it via AppScript. But it was 50% kinda working, so I kept giving the model the output, and it provided updated formulas back and forth for 4.5hrs. Say my time is $100/hr - that’s $450. So even if the new ChatGPT Pro mode isn’t any smarter but is 50% faster, that’s $225 saved just in time alone. It would probably get that formula right in 10min with a few back-and-forth messages, instead of 4.5hrs. Plus, I used about $62 worth of API credits in their not-so-great Playground. I see similar situations of extreme ROI every few days, let alone all the other uses. I’d pay $500/mo, but beyond that, I’d probably just stick with Playground & API.
Re:Brex vs. Mercury – how do they compare? on Tuesday September 17, 2024 @10:42AM (#41568082)
yes, but so much has changed

Brex vs. Mercury – how do they compare?

Brex has a comparison page where they claim superiority in every category, but it's clearly one-sided. I wish Mercury provided a similar page, but they don't. As of September 2024, how do these two actually stack up?For context, we're a funded startup with $300k in cash—below Mercury's $500k threshold for treasury services but acceptable for Brex. Other than that, I love Mercury's UI and...

( Read More... | 2 comments | 2 points | on HN )
Amazing!!! But, jee, think of all the butterflies and other insects picked up by the wind that didn’t make it

Ask HN: A plush toy with local Llama3? Need to outsource kid's "why" questions

I thought of building it myself after seeing the 10 tokens/second that Llama 3 can achieve on an iPhone. However, I'd rather just buy one, lol. I need the Llama 3 to run locally so we can take it anywhere. Any plush toy would do, but a bear or dog is preferred. I just need to outsource my kid's endless "why" questions.

( Read More... | 1 comment | 2 points | on HN )

Ask HN: Is there a tool to extract ALL emails, learn "me", draft responses

I'm looking for a product or startup that can extract all my emails, use GPT-4 to learn "me," then draft responses. It would be nice if it also had other integrations, like Slack, Messenger, Twilio, Todoist, etc., and could extract all those too for a complete second-brain/sidekick/assistant-like replacement of me. I want to grow my side projects, but 80% of my day is just managing day-to-day...

( Read More... | 3 comments | 3 points | on HN )
When could a used Apple Pro Display realistically drop to $3k or less? 12-18 months after first shipments?...