tidbeck

The (closed) beta adds support for Linux.

There was some really good comments by @yurimo on the 'Astra can read and write weights to improve Qwen' hacker news post https://news.ycombinator.com/item?id=50341174

Yes it is (as people already have discussed in this thread). A link to the keynote where Steve Jobs talk about it: https://youtu.be/7zfir0Ide0A?si=fKF68UUoLhXK8cPJ

Zoomed on the image and Imgur now also requires an app…

Forgot about the TTP (stupid me).

I did a small mockup (built by a LLM agent) of a related concept. A page that gets more wear where people interact with it https://app.squidler.io/peggy/patina/

”Serve the model on either two Mac Studios, with M5 Ultra and 512GB of unified memory; or four Mac Studios with 256GB of unified memory, for a total of 1TB of unified memory, enough to hold a NVFP4 quantized version of Muse Spark Open”

Can not find any info on inference speed.

From their blog[1]: "Playground launches today for users in the U.S. (18+) at playground.google, with tiered creation access rolling out based on your Google One membership."

[1]https://blog.google/innovation-and-ai/technology/ai/playgrou...

Should not stores make shopping with AI easier instead, as the number of ”customers” that are agents will increase and they will shop where they can?

Also cache write/read cost + cache efficiency.

It is really hard to find on the site what kvm stands for. Guess people going there knows what it stands for but think it could be spelled out once at least.

I think it also creates music that would have never existed like a song for our beer league hockey team.

Side note: a bit funny they launched ”Muse, a conversational co-writer for ElevenMusic” August 27[1] and Facebook launched ”Muse: The World’s First Personal AI Agent” September 8[2]

Naming is hard!

[1]https://elevenmusic.io/blog/introducing-muse [2]https://about.fb.com/news/2026/09/introducing-muse-personal-...

Seems related but not exactly what was discussed here: https://news.ycombinator.com/item?id=47775633

He took it from a tweet https://x.com/SawyerMerritt/status/1683310992493608960

After signing in I get a 404.

404. That’s an error.

The requested URL was not found on this server. That’s all we know.

I screenshot the info sent from my sons pre-school and send it. I can then ask it about and get reminders in time for odd closing hours etc. Also use it for follow ups on projects, other reminders, research on the go, small apps/games and language training.

(Similar system to OpenClaw)

I agree, learning by doing. Like cook food to get better, do not just read recipes.

OpenRouter entry https://openrouter.ai/stealth/ox-alpha

I would like to have more of a global memory for personal assistant type of agents. Auto updated and maintained. Browsable.

I have been looking for something like this but would actually want to self host due to the potential value/privacy of the data.

What is the difference here from creating a virtual card with Revolut or similar?

Earlier discussion https://news.ycombinator.com/item?id=49364559

Cannot the LLM be seen as a ghostwriter? Have been used by the ones with limited time the resources for a long time. Now we all are in that situation.

May I ask how you solved iMessage and WhatsApp as they do not allow bots as I understand it.

But isn’t that still AI-slop?

That is true but maybe not the thing I want to work on, so much other things to spend my limited time on.

For some of us writing good text takes a lot of time and effort.

So that was just what I was thinking. A lot of text to a explain a fixed seed.

Comparing to Opus 5: Claude Opus 5 (Uncached Input $5/M Cached Input $0.50/M Output $25/M) but you also pay a premium on Cache write 25% for 5m and 100% for 1h.