siwakotisaurav

Thanks for that, looks really good. I can see why they were constantly pushing back fable going out of the max sub with these benchmarks

Yea this one specifically is mostly fable, remotion skills and allowing it to use Gemini 3.1 pro to understand the video

The Claude one was a 3-prompt to get it to recreate the original “Agent view” announcement video. But wanted to challenge it, hence the bankruptcy mode. But it’s really bad at coming up with jokes and a coherent narrative so had to guide it wit prompts like 5 times or so

Fireship video that was just a single prompt, with allowing it to use my basement gpu which has Higgs installed for the voice clone

Now the goalpost is “all” of code?

I don’t think anything I say is likely to change your mind. You might need to get through this valley of depression by yourself in a few years and join the rest of devs using ai

If you take them at their word, the people working on the current browsers, compilers and databases are all using it

Google self reports 70% ai usage in code, bun was fully ai rewritten to be rust

Was about to split my $200 max plan into $100 Claude and $100 codex, let’s see if I still need to

Yea some opinions are just widely shared I guess hn shouldn’t be this political mr … politician

Nice work. Tried remotion when the tweet came out but the results I got were pretty mediocre, some of the results on your homepage are pretty good though.

Saw a recent post about coding on the go and wanted to share a somewhat recent video about doing it on an iPad.

Is a big help for me cause of having to walk a lot due to sciatica

Antitrust in US seems more and more of a joke for every new one.

First few literally broke apart big companies, then we have Microsoft with “just don’t make IE default browser” and now basically nothing for google, forget selling off chrome now, not even banned from making search deals, just “maybe in a few years perhaps reconsider the search deals? Totally non binding tho”

Usually don’t believe the benchmarks but first in web dev arena specifically is crazy. That one has been Claude for so long, which tracks in my experience

You’re using slightly improved qwen and llama not r1. R1 only has 600b model