jack

@theo @NanoBanana new task

@ulintelligence have you read any of my tweets lol

yes

why would I post this otherwise

Currently watching the latest #RevolutionSocial podcast with Taylor Lorenz, and makes me grateful to be posting to my own site (thanks http://micro.blog!), crossposted to my self-hosted Bluesky acco… https://posts.j4ck.xyz/2025/09/04/currently-watching-the-latest-revolutionsocial.html

@ulintelligence No

Why tf waste a button when you can use it for something else

I only just found out what these embedding LLMs do the other day, and Google just dropped a new 308M param model!

I really like how they’re open-weighting(?) these smaller models that can run on super l… https://posts.j4ck.xyz/2025/09/04/google-launched-new-tiny-embedding.html

sell me on silksong please

@thatjoshguy69 Oooh cool

@paytondev @NanoBanana good talk @NanoBanana

@paytondev i’ve thought about this a lot before and it’s the only logical reason i can think of.. companies like Google and OpenAI are very good at consolidating their various models all together into one in their official apps, but the APIs are where things get messy..

@paytondev i assume for APIs? if you’re building an app it’s probably easier to implement an image generation feature with a model that spits out an image 100% of the time, vs one that might spit out text and hallucinate when meaning to generate an image