Bluuue

Developer that loves to learn :D

[Bluuue @bluuueeee01]

@elonmusk no offense but grok is bench maxed like gemini and muse spark... Its not even close to fable and astra and now today with opus 5.5 release it's not even on the same stratosphere.

[Bluuue @bluuueeee01]

@dabit3 I am reviving and old childhood mmorpg that was shutdown over a decade ago over 1 year in and well over 3k hours of decoding proprietary format pushing llms to there limits.

[Bluuue @bluuueeee01]

@bridgebench I hate the benchmaxing when you try the model you know its obvious that these are nowhere near fable and astra.

[Bluuue @bluuueeee01]

@gmi_cloud @Rahul_oog the same wild imagination toddlers have when drawing, thats grok.

[Bluuue @bluuueeee01]

@aksheyd sure looks like a pile of benchmaxed garbage.

[Bluuue @bluuueeee01]

@XFreeze looks like a pile of benchmaxed garbage so far.

[Bluuue @bluuueeee01]

@XFreeze grok 4.7 sure looks like a pile of benchmaxed garbage so far.

[Bluuue @bluuueeee01]

@space_cane @elonmusk @XFreeze So demand is higher then supply in certain dense regions?

[Bluuue @bluuueeee01]

@TechByTaraa Wish I could xD

[Bluuue @bluuueeee01]

@aapakari Haha I love the way you disagree :P To be fair they are not completely useless, they are good for parsing threw data and other tasks that cannot break your production codebase. But from my experience building my emulator they often fail to fix stuff and cause tons of regressions.

[Bluuue @bluuueeee01]

@AreOhEssEyeEe This looks fucking amazing 🤩

[Bluuue @bluuueeee01]

@anah_sahh AI code is easier to understand then most human code.

[Bluuue @bluuueeee01]

@elonmusk @XFreeze Is the demand for bandwidth that much? seems like Starlink is already available mostly everywhere, is it mostly for enterprise? will everyday users experience better uploads and latency? I don't get where all that bandwidth is going, is it for direct to phone service?

[Bluuue @bluuueeee01]

@100rax1 @__roycohen A few days ago I tried 1.3 and it sucked, just like gemini breaking shit at the speed of sound, these models are overconfident in there fixes and cause tons of regressions, at least in complex projects is where SOTA models are much better and spark 1.3 is not even close to fable.

[Bluuue @bluuueeee01]

@100rax1 @__roycohen Muse is unusable similar to gemini, it's benchmaxed garbage, maybe good for parsing data or some useless task cannot be used on production code.

[Bluuue @bluuueeee01]

@elonmusk only one who isnt a complete idiot in this is Ed Zitron.. He calls it like it is, these are glorified chatbots getting overhyped by billionaires to inflate there worth. If it requires 30 million dollars in inference to hack hugging face , it's not something regular people can do.

[Bluuue @bluuueeee01]

@thsottiaux @romainhuet @sama If its not a new model or better limits no one gives a shit, each time you guys say we got something amazing it turns out to be garbage no one uses, people care about 2 things, better SOTA models and better usage...

[Bluuue @bluuueeee01]

@hiarun02 its just you hallucinating

[Bluuue @bluuueeee01]

@stemonteduro yea they are complete garbage, completely unusable right now. I tried a plus plan ran astra on low setting in 20minutes it failed to complete a simple task and ran out of weekly tokens. Make that make sense.. why even offer the 20$ plan?