speedping

Meh. Wake me up when it reasons in latent space and answers in less than a second

Raspberry Pi is going down the way iRobot did, succumbing to pressure from chinese players like radxa or orange pi.

Sad to see industry pioneers lose to players that copy them and undercut prices using cheap labor.

How did this guy get a single-letter GitHub username?

Flaky tests by design is a diabolical idea

So cool. I've fired up pumas (energy monitor) and it seems to run almost fully on the neural engine and not the GPU so it plays really nicely with CoreML

Pre-AI internet data is like pre-war steel

The slop would multiply if we keep feeding it to new models in a loop

I don’t see this as an option in their website

You could theoretically ask “what is the next appropriate character?” and add the entire ascii charset but i doubt it’d work well and you’d be implementing autoregressive churn across network latency…

Probably a beken chip, should be easy enough to flash OpenBeken using the pins if they're exposed You can usually take over tuya devices if you sign up for a developer account and go some sort of verification process, it's pretty easy, sometimes it's possible to get a token for local control (tuya-local)

It basically means that every meaningful comparison produced an embarrassing improvement multiplier

The article's headline contains an em dash. I wonder if it's AI-generated

Their portal gave me a warm feeling of internet nostalgia Also the file claiming to be PDF while actually being DOCX is pretty amusing

I really like the improvements to satelite connectivity, brings us one step closer to a future where LEO broadband like Starlink/OneWeb is a commodity

If OpenCode is so terrible it wouldn't be the base for many of the fortune 500's internal cli coding agents, Meta included

I second that. Gemini 3.5 Flash rocks the benchmark charts but is terrible as an agent. Horrible instruction adherence and makes WAY too many tool calls

Very cool project but having fable in the name makes it sound so tacky

First-party harnesses are great, but i'd really wish this was a CLI and not a GUI

2.764 petabytes of HBM per 8i? So that's where all the RAM went.

I'm working on my IPv9 proposal as we speak. It has an LLM validating the contents of every packet. Gotta stay ahead of the curve.

I can't be the only one who thought this was a phone farm based on iPhones running Apple A16Z SoC (which doesn't exist btw, it's A12Z or A16)

open blogpost → ⌘ + F "pelican" → 0 results ???

This is big for long-horizon tasks

Verification is faster than generation, one forward pass for verification of multiple tokens vs a pass for every new token in generation

> $30/150 MTok Umm no thank you

Looks great! Is it possible to increase the resolution? Each hexagon cover too much area

You articulated it perfectly. I felt exactly what you're describing now while reading the comments to a post about mindfulness a couple of days ago, but I didn't know how to phrase it.

Thank you!