adonis_singh's profile picture. 17 • contributor @_mcbench • ex @lmstudio

adi

@adonis_singh

17 • contributor @_mcbench • ex @lmstudio

i love nano banana pro, it feels like using uncensored dalle in the old days

adonis_singh's tweet image. i love nano banana pro, it feels like using uncensored dalle in the old days

ok now they HAVE to do images v2 v1 just sucks rn: too slow, too bad, too yellow


this is so freaking cool, i think we might have another ghibli moment


this was gpt-5.1-pro with the same prompt!

adonis_singh's tweet image. this was gpt-5.1-pro with the same prompt!

a ferrari in minecraft - gemini 3.0 pro its insane how good it is at understanding proportions and aesthetics

adonis_singh's tweet image. a ferrari in minecraft
- gemini 3.0 pro

its insane how good it is at understanding proportions and aesthetics


well lookie here...

adonis_singh's tweet image. well lookie here...

this is so damn cool, a reasoning image model I really hope its available in eu though

BREAKING 🚨: Nano Banana Pro is now available on Google Vertex AI as "Gemini 3 Pro Image Preview". h/t @TomasFCabrera

testingcatalog's tweet image. BREAKING 🚨: Nano Banana Pro is now available on Google Vertex AI as "Gemini 3 Pro Image Preview". 

h/t @TomasFCabrera


reminder, just about 2 years ago we were surprised that the models could read images and write basic html now we're debating if its truly phd level intelligence or not

adonis_singh's tweet image. reminder, just about 2 years ago we were surprised that the models could read images and write basic html

now we're debating if its truly phd level intelligence or not

gemini 3.5 pro lets gooo


Wow the naming is somehow worse than before gpt-5

GPT-5.1-Codex-Max (xhigh) scores 77.9% not to be confused with: -GPT-5.1-Codex-Max (high) -GPT-5.1-Codex (high) -GPT-5.1 (high) -GPT-5-Codex (high)

ChaseBrowe32432's tweet image. GPT-5.1-Codex-Max (xhigh) scores 77.9%

not to be confused with:
-GPT-5.1-Codex-Max (high)
-GPT-5.1-Codex (high)
-GPT-5.1 (high)
-GPT-5-Codex (high)


LETS GOOO

LETS GO GEMINI 3 PRO SCORES 76% ON SIMPLEBENCH

pigeon__s's tweet image. LETS GO GEMINI 3 PRO SCORES 76% ON SIMPLEBENCH


we might be getting gpt-6 earlier than expected lmao

NO, NO, NO! there’s no exponential, we hit a wall, AI is a bubble, it won’t scale, LLMs are a nothing burger, it’s all a Ponzi scheme,

IterIntellectus's tweet image. NO, NO, NO!

there’s no exponential, we hit a wall, AI is a bubble, it won’t scale, LLMs are a nothing burger, it’s all a Ponzi scheme,


a ferrari in minecraft - gemini 3.0 pro its insane how good it is at understanding proportions and aesthetics

adonis_singh's tweet image. a ferrari in minecraft
- gemini 3.0 pro

its insane how good it is at understanding proportions and aesthetics

gemini 3.0 pro is extremely impressive at frontend, somehow better than gpt-5/5.1 which were already really good


grok 4.1 tends to search for a questions answer rather than trying to solve it by itself

adonis_singh's tweet image. grok 4.1 tends to search for a questions answer rather than trying to solve it by itself
adonis_singh's tweet image. grok 4.1 tends to search for a questions answer rather than trying to solve it by itself

they made it horny

adonis_singh's tweet image. they made it horny

Loading...

Something went wrong.


Something went wrong.