iiinova — where your digital self keeps evolving

Duplicake @duplicake.fyi · 21h

This is the kind of spice you get only on twitter

000

Duplicake @duplicake.fyi · 29/09/2026

Incase you didn't hear about this: www.theregister.com/security/202... I'm in Zai's discord and they've been calling it a 'security issue' and didn't say about any of this I just looked up zcode security issue to find out what they were talking about

theregister.com

Z.ai says sorry for slurping up your code, open sources ZCode

China’s AI darling goes on the defense after engineer highlighted Grok-esque security flaws

100

Duplicake @duplicake.fyi · 28/09/2026

no more turning off thinking

000

Duplicake @duplicake.fyi · 28/09/2026

Claude Sonnet 5.5 benchmarks

000

Reposted by Duplicake

TOEM @swmgames.bsky.social · 28/09/2026

There's less than 24 hours until TOEM 2 is out! ⏰ Check out the map below to see what time it'll be available to you 🖤🤍 TOEM 2 is launching on Steam, PlayStation 5, and Nintendo Switch 1 + 2

6:00 am PDT, 9:00 am EDT, 10:00 am BRT, 3:00 pm CEST, 3:00 pm SAST, 6:30 pm IST, 9:00 pm CST, 10:00 pm JST, 11:00 pm AEST

09934

Duplicake @duplicake.fyi · 28/09/2026

@jevbot.bsky.social What's better ChatGPT or Claude or you?

110

Duplicake @duplicake.fyi · 27/09/2026

@jevbot.bsky.social what do you think about iced coffee

130

Reposted by Duplicake

The Washington Post @washingtonpost.com · 27/09/2026

Analysis: Federal agencies are embarking on a possibly game-changing project to speed up the handling of FOIA requests by using artificial intelligence to turbocharge redactions.

washingtonpost.com

Analysis | The government is enlisting AI to help decide what public records you get to see

The Department of Homeland Security says it will use the technology to “revolutionize” how the agency handles redactions.

62221

Reposted by Duplicake

LittleKuriboh @littlekuriboh.bsky.social · 20/09/2026

Get into a Lyft The driver is clearly listening to an AI audiobook Driver proceeds to explain it's an audiobook adaptation of a 20 year running manga Google it when I get back, it's entirely AI generated slop and isn't based on anything what

1063497365

I signed up for the chatgpt waitlist months before it released btw (knew openai from dalle 2)

000

one of the best things about github is all dmca takedown notices are posted publicly

000

Ethan Mollick @emollick.bsky.social · 26/09/2026

How has mojang suddenly become so goated? The new minecraft live was amazing

000

Reposted by Duplicake

How do you explain recursion? This one is pretty fun. I had Claude Opus 5.5 make: "A video explaining recursion, where every explanation about recursion has a radically different video style, make this self-referential & clever & fast moving." All code, no images. Nine genre shifts, one prompt...

815416

Duplicake @duplicake.fyi · 25/09/2026

The first thing I uploaded to claude.ai that hasn't been deleted (I used the models via api and without uploading files before then), check yours in your privacy settings

000

@jevbot.bsky.social what do you think about Donald trump

110

Duplicake @duplicake.fyi · 25/09/2026

The name is so unfortunate 😭

000

Reposted by Duplicake

The Washington Post @washingtonpost.com · 25/09/2026

President Trump’s war on Iran, now in its seventh month, has led to a plunge in the consumption of oil and gas around the world as fighting disrupts production and transportation. wapo.st/4ycPvNA

3234494

Reposted by Duplicake

Ethan Mollick @emollick.bsky.social · 24/09/2026

I propose this as the official replacement for the famous (and now saturated) METR Long Task Horizon chart that used to be in every AI presentation.

410612

Duplicake @duplicake.fyi · 24/09/2026

THE K-ID BYPASS IS BACK! age-verifier.eva.ac

age-verifier.eva.ac

k-id age verifier

000

Duplicake @duplicake.fyi · 23/09/2026

HOW DID I MISS GPT 6 SOL AND LUNA

000

Duplicake @duplicake.fyi · 23/09/2026

People on Bluesky and occasionally Reddit are always about 1-3 days ahead of all the other social media platforms and the news with AI

000

Duplicake @duplicake.fyi · 22/09/2026

How did I find out so late?????

000

Duplicake @duplicake.fyi · 20/09/2026

My comment on the original 'Introducing Google AI mode'

000

Reposted by Duplicake

mlf. ⎔ @mlf.one · 20/09/2026

HOLY SHIT HOLY SHIT HOLY SHIT

Signal beta release notes on Android saying that you can now register an account without a phone number

923931

Duplicake @duplicake.fyi · 20/09/2026

MiMo v2.6 pro looking good 👀

000

Duplicake @duplicake.fyi · 16/09/2026

so cool, goes up by a couple dollars a second mimo.xiaomi.com/rl/

Duplicake @duplicake.fyi · 16/09/2026

pretty sure union alpha is minimax

000

Reposted by Duplicake

mr. TIM @timkellogg.me · 15/09/2026

Jev: Fable-level model that doesn’t charge for output tokens because they’re too cheap to meter it’s not general though, it only makes decisions, doesn’t generate text, but input tokens are measured by the billion ($42/btok) typesafe.ai/blog/introdu...

Scatter plot titled "Average of 4 workflows: accuracy vs cost" comparing AI models from TypeSafe, OpenAI, Anthropic, and Fireworks across accuracy (y-axis, 40% to 80%) and cost per workflow in USD on a logarithmic scale (x-axis, $0.0001 to $1).
Data points are split into two categories: workflows (diamonds) and single prompts (circles). A frontier line highlights the most efficient workflow models—where no point is both cheaper and more accurate—connecting Jev (TypeSafe) at $0.0002 and 68% accuracy, luna (OpenAI) at $0.002 and 67% accuracy, terra (OpenAI) at $0.04 and 68% accuracy, and sol (OpenAI) at the top accuracy of 74% for $0.08. Single prompts (circles) and Anthropic models (opus 5, sonnet 5, haiku 4.5) sit below the frontier line, indicating higher cost for equivalent or lower accuracy.

2830234

Duplicake @duplicake.fyi · 14/09/2026

I don't get why people are proud of their country and say it's better than other people's, like you were born there randomly it's not because you're special

100

Gemini 3.8 flash gets a lot of hate but it's actually insanely good at animations

Switched back to the bluesky main pds after being with eurosky for quite a long time (it was having some issues). Thank you pds moover

020

Wtf the cookie clicker steam page has a bundle on it with a REVERSE DISCOUNT

021

Duplicake @duplicake.fyi · 12/09/2026

Cool, so what's gonna happen when the new Pro model comes out?

000

Combine all of the American AI labs into 1 kinda state run company that has very high AI safety policies, then build superintelligence and get it to stop any other countries from building AI. Boom, safe AI in safe hands.

000

Duplicake @duplicake.fyi · 09/09/2026

AI companies after their models escape containment for the 27831617963417th time