jon3k

healthcare cio, rhce and 少し日本語

Disclaimer: views are my own and not necessarily representative of my employer.

[jon3k @jon3k]

"How's AI going?"

Me:

[jon3k @jon3k]

@InvestInJapan This is my favorite burger in Tokyo. It's near Yoyogi Park.

[jon3k @jon3k]

@dashcammusic I want to feed all these images into AI and tell it to make me a 1980s mystery/thriller game set in Miami. Such a vibe.

[jon3k @jon3k]

@scoutingdad2022 @ammolab1 .22 is 2-3 MOA at 300 yards. The cartridge itself is not capable of what you're describing.

https://t.co/0hKyEhZXgZ

[jon3k @jon3k]

@AlastairGrayson @malonepulse1 I assume you're joking but this is fun to think about.

Would take over three hours for a single person to download 210GB via OC3 😂

It would take about 350 years for 1M people to download 210GB via OC3.

[jon3k @jon3k]

@initjean @StijnSmits Argon and Carbon are both fantastic and usage limits have been great on Ultra. Cranking away with the new swarm mode for hours with a mix of Argon workers and Carbon orchestrator/reviewer. Extremely impressed so far.

[jon3k @jon3k]

@cassiothird @thdxr It's available right now.

[jon3k @jon3k]

@ThePrimeagen Models are good enough for most work. All that I want now is 0.1 second time to first token and 1k tokens/sec.

[jon3k @jon3k]

@s_torian_ Labs are massively subsidizing AI at a loss. We should all be taking advantage of it.

[jon3k @jon3k]

@alexmichael [x] doubt

[jon3k @jon3k]

Did they finally find a way to measure benchmaxxing? That would be amazing.

[Justin Lo @loresjpg]

Nice DeepSWE score! Does it survive a lowercase filename?

Introducing MaxxBench: 89 DeepSWE tasks with small changes to the instructions or starting code. Same engineering problem, slightly different requirement.

Models clustered around 74% on DeepSWE end up 23 pp apart.

Good at coding, or good at DeepSWE? 🧵

[jon3k @jon3k]

@Dataphil3 @coffeecup2020 No. They are releasing four Qwen models. But yes one will be a 30b class model.

[jon3k @jon3k]

@thenickmoy It's been the default pattern for all of us too using the other labs 😂

[jon3k @jon3k]

@brockpierson Remind me bot

[jon3k @jon3k]

Is the bubble bursting or just a small post midterm correction? I guess the next week or two will be very telling.

[Polymarket @Polymarket]

BREAKING: Krispy Kreme officially outperforms NVIDIA over the past year.

[jon3k @jon3k]

@justabdulraouf @justalexoki It's not weird at all if you speak Japanese and read the original tweet in japanese. It's actually a really great pun.

[jon3k @jon3k]

@NonoLazyRobot @DesignArena @thinkymachines @Kimi_Moonshot They've released weights for their other models. Same thing, release it, commit to a date, release as Apache 2.0. I expect the same here.

[jon3k @jon3k]

This seems insane? This is like a 600B MoE US domestic open weight (soon!) model.

[DesignArena @DesignArena]

BREAKING: Inkling 2 Small by @thinkymachines takes 4th place overall on DesignArena and is 1st place among open weights models with an Elo of 1376!

At $0.30/$1.00 per 1M input/output tokens, Inkling 2 Small is 15× cheaper on output than Kimi K3 by @Kimi_Moonshot. This model also places 1st on 2 of our subcategories including General Purpose and Landing Page prompts. Model weights to be released soon.

A huge congrats to the @thinkymachines team on a top-5 placement on the leaderboard!

[jon3k @jon3k]

@MKUltraMoney I go to Japan a few times a year and it always takes a few weeks before twitter stops giving me japanese ads 😂

[jon3k @jon3k]

@cheeez42 @Beelinkofficial I also made the best remote app ever imo

[jon3k @jon3k]

@cheeez42 @Beelinkofficial I bought three in the last month for testing a little app Claude built for me. I love them.
https://t.co/q8VhEAK1OV