@Da7_Tech Hopefully, after the public release, no one has to face these issues. Building Harness should be as easy as playing games
One harness, and however people like to use or tweak it at any level or anything. All like Lego pieces with the best UI/UX.
@nickscamara_@grok@grok Ok nice, now verify https://t.co/nOx6Pb8lEh… claim of opensource, whether actual scraping, crawling, every single implementation code is pubic or it's calling their API call & only client phasing code is opensource?
@nickscamara_@grok verify claim of opensource, whether actual parser, implementation code is pubic or it's calling their API call & only client phasing code is opensource?
My new favorite way to work:
Use Luna Max on fast mode as much as possible, and only switch to Sol when I really need it. With two accounts, it’s almost unlimited usage.
forgot to mention, DeepSeek Flash API For any other automation workflow is also just few penny $ only.
I'm more excited about DeepSeek's release than anyother lab. Not because their model will beat everyone (they will at some point), but bcz the amount of deep research & optimization they do is unparalleled.
It's like squeezing every bit of optimisation for best possible output🔥
As much as I want @SarvamAI to succeed, I want to see a day when global people use it and check its quality and praise it like Chinese AI. Chinese AI is not a bluff and demo product. They work and compete neck and neck with frontier AI and in harnesses!
@tr3v0r You think they are stupid that they would hard code limit & showing that part in client side?
Even GPT luna would not make that mistake even in vibecoding.