scumfuck flowergirl // i know a little bit about computers // priv: @Zbuildstd
17
Joined December 2020
- Tweets20.6K
- Following1.8K
- Followers5.9K
- Likes55.2K
nullptr π±π© retweeted
me showing Santa my Digital ID (securely encrypted by the number 4) after he falls down my chimney on christmas eve
up next, labour sets out to ban:
- bicycles (what if they fall off)
- christmas (santa is a bad role model; fat)
- water (what if they choke)
- leisure (could promote "fun" to impressionable 16 year old babies)
nullptr π±π© retweeted
get a life if you're crying about the Big Walk devs using LLM help for them to finish the game
everyone's asking for the wallpaper: it's "plasmic" from my friend @z2rMC's wallpaper pack. it's free on his gumroad, you should go there to download it instead of getting it from this image liamz2r.gumroad.com/l/Wallpaβ¦
nullptr π±π© retweeted
Introducing worse_query, a zero-day sandbox escape for iOS 26 and 27, again.
In collaboration with @0xjohnny!
github.com/forcequitOS/worseβ¦
i feel gpt-3.5-turbo should be open-weights or at least preserved via api. they stopped serving the last checkpoint a few weeks ago, but it feels like too historic of a model to be left behind
nullptr π±π© retweeted
Qwhen? Qsoon!
For our friends outside China, the Chinese text in the poster comes from Zhuangzi, an ancient Chinese classic:
"Far in the northern sea lives a fish named Kun (pronounced like Qwen). So immense is its size that no one knows how many thousands of miles it spans."
when china do it: benchmaxxing
when the west do it: βspecializationβ
Today, we're announcing webAI Meridian-1 CivOE.
A 12B AI model that statistically matched Gemini 4 Argon and outperformed Claude Opus 5.5 in evaluation.
But the most important breakthrough isn't the benchmark.
It's how we achieved it.
Instead of endlessly scaling model size, we've built intelligence that adapts to the work through specialized experts.
Our results:
Meridian-1 CivOE: 85.5
Gemini 4: 84.7
Claude Opus 5.5: 82.7
GPT-5.6: 71.0
The same 12B foundation went from 59.8 to 85.5 through specialization. A 25.7-point improvement without making the foundation model larger.
We believe this represents a fundamental shift in how intelligence will evolve.
Just as software became personal, customizable, and accessible on everyday computers, intelligence is heading in the same direction.
Intelligence that adapts to your knowledge, your work, and your devices.
We're beginning the rollout of Meridian-1 CivOE with strategic partners over the next few weeks, before expanding access to all webAI users.
The future of intelligence isn't just bigger.
It's yours.
Read More: webai.com/research/meridian-β¦
nullptr π±π© retweeted
absolute pangram mandate, more good news
EXCLUSIVE: Today, Pangram is releasing a new model that can detect AI-generated video.
Itβs one of two new releases, including an updated static image detection model. Per Pangram, the new image detection model achieved 99.96% accuracy on an internal benchmark; video detection scored 98.57% accuracy.
Said CEO @max_spero_: βWeβve always felt that itβs going to be important to know whether somethingβs AI generated or not and let people make their own decisions.β
@evanmilenko reports π
nullptr π±π© retweeted
>"outperforms american open weight models"
>its closed weights
oh okay to be clear iβm talking about speeds. intelligence really seems to be the same lol but i canβt bear to use it when it runs at anywhere between 4 and 50 tk/s, especially when deepseek serve it at like 240tk/s consistently
