I had the initial idea for the ALLEIKHREOS intro while camping in the wilderness without a computer.
I suddenly woke up at 3am with a full track in my head, and had to run to the bathroom (as to not wake my friends) and use my voice memos to get the idea down. Listening to me trying to explain it to my future self is so funny 😂
A few months ago @ymirymirymirymir and I got the massive opportunity to score the entire final boss battle for Arknights: Endfield.
It was a crazy few months, but I’m super proud of what we created 🤍
Mobile vs PC Graphics Comparison!
Mobile has never been this close to PC ever before, even the Mobile Graphics look absolutely stunning..
At some points I thought wait this looks exactly like PC 😭
C : BleeeGMV
How Indonesian “justice” works:
—Arrest Nadiem with no evidence of wrongdoing as he tried to reform the country as a minister.
—Prosecutor loses face in court due to the ridiculousness of the case.
—To save face, the judge gives Nadiem 10 years in jail anyway.
—Next step? Nadiem is likely released on a presidential intervention.
Indonesia is a laughing stock on the global stage.
My entire AI stack is now Chinese 🇨🇳
87% cheaper. same revenue
swaps by task:
1. reasoning / backend brain
Opus 4.8 → Kimi K2.7
benchmark gap: ~8% · price: ~11x cheaper
2. code generation
GPT-5.5 → Qwen 3.7 Max
benchmark gap: ~18% · price: ~7x cheaper
3. agent loops + tool calling
Sonnet 4.7 → GLM 5.2
benchmark gap: ~3% · price: ~5x cheaper on input
4. cheap volume / bulk processing
GPT-5.5 mini → MiMo V2.5
benchmark gap: ~6% · price: ~12x cheaper
5. image generation
GPT-Image-2 → Wan 2.5
benchmark gap: ~5% · price: ~8x cheaper
6. video generation
Sora 2 → Kling 3.0
benchmark gap: roughly equal · price: ~6x cheaper
[ result after 30 days: ]
operating costs dropped 87%, output quality dropped 4% on average, revenue unchanged
the most important that these models will be not banned in a month and i can run them locally
nobody will steal my data and i can learn them as i need
full article drops tomorrow with:
> exact routing logic per task type
> the 2 cases where I still pay for American
> the migration playbook anyone can copy in a weekend
VERY IMPORTANT to get migrated now, while it's not too late
Yang motong namanya bukan "biaya PLN".
Namanya PPJ. Pajak Penerangan Jalan.
Iya, lo gak salah baca. Pajak.
Tiap kali lo beli token, lo bayar pajak.
Dan ini yang bikin gw paling kaget: duitnya gak masuk ke PLN sepeser pun.
PLN cuma kasirnya. Duitnya disetor ke pemerintah daerah lo.
Temen gw kerja di PLN.
Dia bilang ada satu pertanyaan yang tiap hari bikin dia dimaki, dan dia capek ngejelasinnya.
Pertanyaannya: "kok beli token 100 ribu, dapet listriknya kurang? Sisanya ke mana?"
Dia cerita ke gw sambil geleng-geleng. "Yang motong duit itu bukan PLN. Tapi yang dimaki tetep kita."
Gw ngerutin dahi. Kalo bukan PLN, terus siapa yang ngambil?
Malem itu gw cek struk token gw sendiri. Ada satu baris kecil yang selama ini gak pernah gw baca.
Andrej Karpathy spent 2h showing how he actually uses AI day to day
he's a co-founder of OpenAI and led AI at Tesla, so when he shows how he works, it��s worth watching
and the whole session is just him telling the machine what he wants in simple terms, like he's briefing a coworker
watch what's actually happening the entire time:
> he describes the task in normal words
> it goes off and does the work
> he glances at the result and nudges it with one more sentence
that's the whole skill, and you've had it since you learned to talk
the only gap between that and a worker that runs on its own is handing that sentence a schedule and the tools to act
check his work, then build the version that keeps working when you stop
Introducing Qwen3.6-Max-Preview, early preview of our next flagship model
·Improved agentic coding capability over Qwen3.6-Plus
·Stronger world knowledge and instruction following
·Improved real-world agent and knowledge reliability performance
Smarter, sharper, still evolving.