@Nltducf2 @ConflictTR "Hiç tanımadıkları, bilmedikleri bir insana karşı şöyle bir nefret kusabilmek gerçekten iğrenç. Sadece dininden dolayı ha"
Bu yazı bana yahudileri hatırlattı :d
Türkçe
Can
174 posts
@ConflictTR Hiç tanımadıkları, bilmedikleri bir insana karşı şöyle bir nefret kusabilmek gerçekten iğrenç. Sadece dininden dolayı ha...
Türkçe
@clanka9000 @Pirat_Nation "He gave OpenAI 20M"
Lmao is this supposed to be a bad thing?
You said, RAM prices are high because he was a part of it by donating. In reality RAM prices still would've raised even though he donated or not lmao. Fuck outta here with bs propaganda.
English
@CA1337s @Pirat_Nation now tell how that changes the fact that he gave OpenAI 20M
English
In 2018, Gabe Newell quietly helped OpenAI a lot, he gave OpenAI $20 million because he believed early on that AI should grow in a safe and smart way
At that time, OpenAI was still deciding its future plans.
Newell was the only member of a small, unofficial advisory group and gave them advice during that key early stage.
English
@clanka9000 @Pirat_Nation >funds openAI
>RAM prices gets fucked hy AI
Are you stupid? Google was gonna do it if it wasn't OpenAI
English
@Pirat_Nation >funds openAI
>RAM prices gets fucked hy AI
>every single gaming hardware gets a price increase
>controller with no Xinput is $100
>steam machine gets internally delayed
LMAOOOO CATTLES GET WHAT CATTLES DESERVE
English
@thsottiaux @thsottiaux But will it actually know the work is finished fully? Or it will still need a human interfere with it?
English
You can now keep codex going for days.
With GPT-5.5 it will build an entire OS kernel for you if you ask, or find critical bugs in a codebase, or optimize your database schemas, or… the options are endless.
Felipe Coury 🦀@fcoury
/goal also lands in Codex CLI 0.128.0. Our take on the Ralph loop: keep a goal alive across turns. Don't stop until it's achieved. Built by my co-worker and OpenAI mentor Eric Traut, aka the Pyright guy. One of the GOATs I get to work with daily.
English
@eepromblaster @TayfunKOfficial PAL NTSC alakası yok.
Self Control çalıyorsa 2.00 veya 3.00 versiyonunu Billie Jeans çalıyorsa 1.40
Türkçe
@TayfunKOfficial Self Control çalıyorsa PS2 NTSC sürümünü oynuyorsunuzdur Billie Jean çalıyorsa PAL sürümü oynuyorsunuzdur denendi onaylandı isteyene dc de gösteririm
Türkçe
GTA Vice City'de oyuna başlarken motora binerseniz bu şarkı, araca binerseniz Self Control çalar...
Alın bu bilgiyi ne yapıyorsanız yapın...
Siz yine de GTA Türk City modunun "Seninle İlk Defa" şarkısını dinleyin!
Hayden's Song Of The Day@HaydenSOTD
SOTD Day 28: Billie Jean - Michael Jackson (Number 438)
Türkçe
@_architected @googlegemma "I am not your man."
"And yet you’re angry"
Hey buddy you're not on Reddit.
English
@CA1337s @googlegemma I am not your man.
> there's no smart AI that will run on any consumer device anyway
And yet you’re angry that they shown some simple game as an example. Meh.
English
Offline vibe coding with Gemma 4.
This open-source Electron app runs entirely on your Mac via Apple's MLX framework.
No Wi-Fi required.
English
@_architected @googlegemma I didn't said it would run perfect neither. Just bare minimum. You said "it won’t run on like 90% consumer device"
My man, there's no smart AI that will run on any consumer device anyway. You asked for Gemma alternative and it's Qwen and it's better than Gemma.
English
@CA1337s @googlegemma You know that 27b non MoE won't run perfect on 16 gigs, right? And at the same time you expect it to be smart lmao
English
@_architected @googlegemma My man if you tryna run a local LLM with less than even 16GB VRAM you shouldn't be doing that at all lmao.
English
@CA1337s @googlegemma Yeah and it won’t run on like 90% of the consumers devices due to 27b size
English
@Faaaaaaaaahhh @ntahabercom "Inbred cuckold piece of pedo worshiper"
I'm not Indian saar
English
@CA1337s @ntahabercom Go eat cock you inbred cuckold piece of pedo worshiper shit
English
Hindistan’da bir sokak şekercisinin “nano teknolojik eldivenleri” sosyal medyada merak uyandırdı. Görüntülerde hijyen seviyesinin adeta uzay standardında olduğu görülürken, çocukların “temassız üretimden” çıkan şekerleri yemek için sıraya girdiği anlar dikkat çekti.
Türkçe
@CA1337s @ntahabercom No, they pedo worshiper just like you fucktard
English
@ntahabercom I think you are blind this is Pakistan and not India you looser
English
@scaling01 I really don't care about benchmarks. Why? Look at Gemini 3.1 and it's benchmarks. It cannot even do the simple task i'm throwing at it.
I didn't said it's GPT 5.4 level i said something between 5.2 and 5.4
English
please show me the >6 hour METR time horizons, WeirdqML, GSO, PPBench, LisanBench, SimpleBench, RLI or ARC-AGI-2 scores if you are so confident that open-source models are at GPT-5.4 or Opus 4.5+ level
there's still a big gap and it's likely getting bigger
Can@CA1337s
@scaling01 It can do what GPT 5.2 can do. In my own tests it also did some tests GPT 5.4 failed to do so. It's between 5.2 and 5.4
English
@thsottiaux @thsottiaux
- Subagents are not really effective. I wish they worked similar like in Claude Code
- UI really need big improvements... All that tool calling like Ran, Grep is just bricks of text to my eyes
- Please add a fricking /undo so we don't waste our tokens to undo.
English
@scaling01 It can do what GPT 5.2 can do. In my own tests it also did some tests GPT 5.4 failed to do so. It's between 5.2 and 5.4
English
I do think Kimi, DeepSeek, GLM and ByteDance are close to ~Opus 4.5/GPT-5.2 reliability or will be there in the next 1-2 months
so they are reaching escape velocity too
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex
They say "12 hours" or "13 hours" but of course wall clock time has nothing to do with METR time horizons which are established relative to human labor. Still I wonder: have Chinese models finally reached the SWE takeoff stage? 3.6-Max, 5.1, 2.6. It does feel like a step change.
English
Kimi K2.6 seems to have been released on their website.
The Pelican SVG turned out super impressive. I’m surprised, I’ll test it more.
AiBattle@AiBattle_
Kimi K2.6 seems to be live on the website now
English
@diegocabezas01 That and illusion of user having thinking next model is even more powerful than previous one
English
@CA1337s Do you think is on purpose or because they are reallocating resources on the back end?
English