ntcho

10 karmajoined 5 tahun yang lalu

Submissions

OpenCode Data: Real-world AI model usage, cache ratios, and costs

2 pointsby ntcho3 hari yang lalu1 comments

comments

3 hari yang laludiscuss

Came across this dashboard tracking live LLM usage on OpenCode users. Interesting production metrics, including token volume, session costs, and cache hit ratios. DeepSeek v4 currently dominates their market share.

ntcho

11 hari yang laludiscuss

Gosh, the last sentence is the most terrifying thing I’ve read today.

ntcho

bulan laludiscuss

Average Firefox experience right here

ntcho

2 bulan yang laludiscuss

absolutely love the domain here. great taste

ntcho

2 bulan yang laludiscuss

I landed on a similar workflow. I found letting agents navigate / self-update docs on a repo full of markdown files quite successful

ntcho

2 bulan yang laludiscuss

username checks out

ntcho

2 bulan yang laludiscuss

Have you tried MacsFanControl? MBP fans rarely turn on by themselves, but controlling yourself makes throttling not a problem

ntcho

3 bulan yang laludiscuss

I’m also having a blast with generating userscripts/css to make web SaaS work better for my workflows. Went from waiting for months for a new feature to writing QoL improvesments myself

ntcho

4 bulan yang laludiscuss

I also noticed Gemini web UI failing to send messages on Firefox. Anyone else?

ntcho

6 bulan yang laludiscuss

I think this should be on HN hall of fame: https://hn.algolia.com/?q=sabotage+manual

ntcho

8 bulan yang laludiscuss

Decades of progress made in hardwares and networks… just to be negated with 100kb+ webpages

ntcho

9 bulan yang laludiscuss

This is hilarious, thanks for sharing. Kinda crazy how well it works and already better than some ads

ntcho

9 bulan yang laludiscuss

Has some quirks, but Beeper is exactly what you are describing: https://www.beeper.com/

ntcho

9 bulan yang laludiscuss

This reminds me of cleaning a toaster with a dishwasher: https://news.ycombinator.com/item?id=41235662