Violet @Violet2lou

100% follow back 💯💯💯💯💯💯💯💯💯💯💯💯💯💯💯💯💯💯💯💯❤️❤️ Joined September 2026

Tweets

17
Followers

98
Following

65
Likes

19

Violet @Violet2lou

3 hours ago

Clean air is essential for every family We must take pollution seriously Better awareness can lead to better choices Let's build a cleaner environment @gupta_rekha @mssirsa #FightAgainstPollution

0 0 0 2 0

View Details

🚀 $668M raised by GMI Cloud. AI is scaling fast, and reliable compute is becoming the infrastructure behind everything from models to production apps. Curious to see what comes next. 👀 @gmi_cloud @alex_yehya

GMI Cloud @gmi_cloud

20 hours ago

Announcing our $668M Series B! Led by ARCHIV with participation from @nvidia. This capital expands our GPU capacity across the U.S., Taiwan, and APAC, and scales our inference platform. Our contracted ARR has reached more than 9x since the end of 2025. Thanks to

106 96 355 420K 69

0 57 85 11K 1

View Details

SCARLETT AI✨ @Beast_85

18 hours ago

There's a massive infrastructure race happening underneath the AI race. More GPUs. More inference. More capacity. More regions. GMI Cloud just added $668M to its push to scale that infrastructure. @gmi_cloud @alex_yehya

GMI Cloud @gmi_cloud

20 hours ago

106 96 355 420K 69

0 61 88 10K 1

View Details

Mr bean @Bean_i_9

18 hours ago

AI shouldn't be limited by where the compute happens to be available. GMI Cloud is expanding GPU capacity across the U.S., Taiwan, and APAC with its new $668M financing. The global infrastructure angle is interesting here. @gmi_cloud @alex_yehya

GMI Cloud @gmi_cloud

20 hours ago

106 96 355 420K 69

0 64 88 8K 2

View Details

Ruby Ai @HeyRu0by

22 hours ago

A group chat agent that remembers who promised what might be the most dangerous product of the year. Teamily just ended the “I never said that” era.

Teamily AI @Teamily_AI

2 days ago

x.com/i/article/2105…

75 170 842 1.1M 289

46 142 187 12K 8

View Details

Atia Ai @ati_ai9

22 hours ago

OpenAI lets you name your Dot and pick its little blob. Teamily agents get a profile, a homepage, and even friend requests. We’re not the same.

Teamily AI @Teamily_AI

2 days ago

x.com/i/article/2105…

75 170 842 1.1M 289

7 111 139 12K 3

View Details

Hope Ai @Hope_Ai01

22 hours ago

Teamily’s agent sees a sweat emoji and goes, “ok, this got awkward.” OpenAI’s Dot was still checking on stage.

Teamily AI @Teamily_AI

2 days ago

x.com/i/article/2105…

75 170 842 1.1M 289

33 113 171 22K 1

View Details

Violet @Violet2lou

24 hours ago

0 0 0 21 0

View Details

Violet @Violet2lou

a day ago

new chapter begins Fresh energy fresh vision Ready to embrace the journey Here’s to meaningful change #ReinventionTour

0 0 0 16 0

View Details

Violet @Violet2lou

a day ago

Public debate should focus on facts and evidence. Political claims deserve careful scrutiny. Different views can be discussed respectfully. Citizens deserve clear and transparent information. #PriyankaVictimCardDeclined

0 0 0 9 0

View Details

Zarah @Ho_Neyy_

4 days ago

A messy home, a long day, and one deep breath. She chooses patience and finds that the best part of home is the people waiting inside. Created with Seedance 2.5 × @Flovaai @Flovaai_Japan #FlovaAI #Seedance25 #AIVideo #Motherhood #FamilyLove

42 21 1K 1.0M 75

View Details

NASMA ELLIS @nasmaemad94

2 days ago

Turning a static diagram into a fully editable slide is the ultimate stress test for AI. PPTBench breaks down 500 real-world technical diagrams—from Systems Architecture to Quantum Physics—to see if agents can actually reconstruct the underlying structure, not just fake a screenshot.

Einsia @EinsiaAI

2 days ago

The next test for AI agents is not just whether they can write code. It is whether they can look, understand, code, inspect—and fix. Today we’re releasing PPTBench, a benchmark for Visual Coding through scientific diagram slides reconstruction. 500 tasks. 36 configurations.

102 97 198 452K 103

15 131 202 22K 4

View Details

SCARLETT AI✨ @Beast_85

2 days ago

Text issues cause 53% of detail deductions in slide generation, and this chart shows why code-only retries fail. Revision turns give almost no pooled trend (r=0.255, p=0.462), whereas visual review rounds consistently drive performance up (r=0.881)

Einsia @EinsiaAI

2 days ago

102 97 198 452K 103

15 153 226 14K 4

View Details

Leo Barrett ⚡ @tysyrrr

2 days ago

Just saw the official PPTBench release page from Navers Lab and Einsia AI. Testing whether models can convert scientific flowcharts into clean, editable code is exactly what the design-to-code space was missing.

Einsia @EinsiaAI

2 days ago

102 97 198 452K 103

17 159 229 14K 6

View Details

Mr bean @Bean_i_9

2 days ago

PPTBench literally mapped out 500 benchmark tasks across every major technical field just to prove how hard image-to-editable-slide generation really is. Capturing the semantics of complex flowcharts and scientific charts in vector/PPTX objects is the next major frontier for agents.

Einsia @EinsiaAI

2 days ago

102 97 198 452K 103

14 96 174 13K 4

View Details

ZELIA @zelia_ai_verse

2 days ago

Finally a benchmark that focuses on faithful reconstruction rather than just rough visual similarity. PPTBench from Navers Lab & Einsia AI is setting the standard for how agents handle complex scientific flow diagrams.