D🪐PE

[D🪐PE @D1G1T4LD0P3]

@LyalinDotCom It is exceptionally smarter and faster than 3.8 flash and that was only released a bit ago. Great progress! enjoying it w/ ultra

[D🪐PE @D1G1T4LD0P3]

@OfficialLoganK AMAZING!

[D🪐PE @D1G1T4LD0P3]

@nvidianewsroom Crazy how they let China win on this.

[D🪐PE @D1G1T4LD0P3]

@GoogleAIStudio @OfficialLoganK Can we get technical clarity on AI Studio limits for Google AI Ultra? Ultra says "5x higher than Pro", but Pro has zero AI Studio perks listed. What are the actual RPM/TPM/RPD thresholds vs Pay-As-You-Go Tiers? Devs need concrete metrics for local build loops.

[D🪐PE @D1G1T4LD0P3]

@googledevs I cant even test anything in AI studio anymore. Whats up with this?

@OfficialLoganK

[D🪐PE @D1G1T4LD0P3]

@NickRokke Its their fault for making it so inefficient. its embarrassing how much more 3.8 and 4.0 takes compared to 3.7 high for comparable tasks

[D🪐PE @D1G1T4LD0P3]

@_philschmid The frustration is 3.7 Flash was deprecated after 2 months despite hitting the sweet spot for speed and token efficiency, while 3.6/3.8 stuck around. Jumping to heavy architectures like Argon without keeping that lean Flash tier disrupts real workflows.

[D🪐PE @D1G1T4LD0P3]

@_philschmid I'll keep reiterating this even if its on deaf ears:

We need to go back to the 3.7 Flash approach - light token usage, fast, and smart at high thinking levels. Argon feels heavy and expensive for no reason, and to make things worse it's still confused about what year we're in

[D🪐PE @D1G1T4LD0P3]

@haider1 They are selling their TPU capacity to Anthropic

[D🪐PE @D1G1T4LD0P3]

@googledevs We need to go back to the 3.7 Flash approach - light token usage, fast, and smart at high thinking levels. Argon feels heavy and expensive for no reason, and to make things worse it's still confused about what year we're in

[D🪐PE @D1G1T4LD0P3]

@LyalinDotCom today must be captain obvious day from all the google employees.

lmao

[D🪐PE @D1G1T4LD0P3]

@elshayib_ Google: in the mean time let me remove some modals so you got less to use

[D🪐PE @D1G1T4LD0P3]

@GoogleAIStudio @OfficialLoganK
bring back 3.7 flash????

[D🪐PE @D1G1T4LD0P3]

@SalmanNaseer @antigravity Totally agree, 3.7 is so much better than 3.8. what a shame. really annoyed

[D🪐PE @D1G1T4LD0P3]

@alwayspriyesh This has to be because everyone was using 3.7 over 3.8

[D🪐PE @D1G1T4LD0P3]

@OfficialLoganK no troll I got to be one of the top .01 percent of power users for AI studio and 3.8 is literally god awful. 3.8 high thinks too long doesnt actually do anything useful. 3.8 medium is pretty stupid

3.7 high > greater than anything I've ever used for serious large context tasks

[D🪐PE @D1G1T4LD0P3]

@thtbee_ Google is destroying their AI or realized no one uses 3.8 because its ass and they need people to use it over 3.7 for testing i assume. 3.8 is fucking awful

[D🪐PE @D1G1T4LD0P3]

@OfficialLoganK nah this is crazy. 3.7 flash was / is so much better

[D🪐PE @D1G1T4LD0P3]

@OfficialLoganK What happen to Gemini 3.7 flash in AI studio? Not sure where else to post this but its no longer available

[D🪐PE @D1G1T4LD0P3]

@ChanduThota AWESOME CHANGE!!! ive been having to convert MD to DOC with external tools for a while now. thanks!