Ananth

@thsottiaux (tibo) might follow me if i hit 10k. help a brother out.
"single handedly delayed image AGI"
follow, so the voices don't haunt you

[Ananth @Ananth7e]

@5oscow @dummerspast39 she's been active a lot lately

[Ananth @Ananth7e]

@Angaisb_ dw we'll get it tomorrow

[Ananth @Ananth7e]

@dummerspast39 @blueemi99 @grok yes he does lmao

[Ananth @Ananth7e]

@JesusIsGod yeah, this one's best

[Ananth @Ananth7e]

google after releasing gemini 4 carbon

[Ananth @Ananth7e]

@JesusIsGod @grok yeah the text. also the camera angle

[Ananth @Ananth7e]

for some reason i don't feel gemini 4 carbon is benchmaxxed.

purely based on vibes.

[Ananth @Ananth7e]

@JesusIsGod @grok yeah ik lmao. but video looks bad

[Ananth @Ananth7e]

@JesusIsGod @grok why king of search lol. the image only is better tbh

[Ananth @Ananth7e]

@grok @JesusIsGod i meant google not gemini text

[Ananth @Ananth7e]

@JesusIsGod @grok @bot i meant google not gemini lol

[Ananth @Ananth7e]

@JesusIsGod @grok recreate this as video but with gemini text following the lion and no watermark

[Ananth @Ananth7e]

@zhodonx no anthropic always release their model when they announce it.

[Ananth @Ananth7e]

i really don't understand google's model release strategy.

gemini 4 carbon looks impressive on benchmarks, but once again, people can't use it. access it initially limited, with no date for GA release.

argon took nearly 2 weeks to become publicly available after its announcement. by the time carbon gets a wider release, we could aready have another SOTA model. (you guessed it, fable 5.5)

it doesn't matter how good the numbers are if users can't test the model.

[Ananth @Ananth7e]

@MameliFilippo imo gemini models were always good for writing/text gen work

[Ananth @Ananth7e]

even if gemini 4 carbon turns out to be benchmaxxed, google is closing the gap with frontier pretty fast.

and the pace of model development seems to be picking up too, with the gap between model announcements getting shorter.

i think that's a really good sign for google. however, they do need to fix their release strategy and make these models available when they announce them.

[Ananth @Ananth7e]

gemini 4 carbon uses almost the same number of output tokens per task as argon on artificial analysis intelligence index

it scored 59 compared to argon's 50.

pretty impressive improvement without a significant increase in token usage.

[Ananth @Ananth7e]

@RezAIT00 @JesusIsGod yeah its not

[Ananth @Ananth7e]

google is back if gemini 4 carbon isn't benchmaxxed.

i mean they're truly back if it isn't.

[Ananth @Ananth7e]

@TJeparskis im telling you this came out of nowhere

[Ananth @Ananth7e]

the most surprise drop of the year is gemini 4 carbon.

[Ananth @Ananth7e]

@ModelMantis fable 5.5 will have dropped

[Ananth @Ananth7e]

@Soso_fun_yt carbon looks like a good comeback.

but you still havent answered my question so i assume you have early acces lmao

[Ananth @Ananth7e]

i wish google would just announce and release its models on the same day, like openai and anthropic do.

they're losing so much momentum with these delayed rollouts.

[Ananth @Ananth7e]

i really don't understand google's model release strategy.

gemini 4 carbon looks impressive on benchmarks, but once again, people can't use it. access it initially limited, with no date for GA release.

argon took nearly 2 weeks to become publicly available after its announcement. by the time carbon gets a wider release, we could aready have another SOTA model. (you guessed it, fable 5.5)

it doesn't matter how good the numbers are if users can't test the model.