israrkhan

This is nice

so this is what pacing down looks like

I have not followed up x86 emulation recently, but is it a new project? or a port of bochs for Mac?

Some of the charts are not uptodate.

For example "Raw capability: Math Index", shows GPT-5.2 as highest. It does not even mention astra, of fable.

Excellent resource for beginners

Models are next word predictors. Without tools and execution environments they cannot perform any action. Typically the tools are running in a seperate systems and sandboxes that do not have access to where model weights are hosted.

IMO This type of exfiltration can happen only from locally running models (which are perhaps already opensourced models), not from a frontier lab.

This is a crowded space with lots of solutions (oss and commercial)

https://github.com/anderspitman/awesome-tunneling

For someone looking for an opensource solutions, following is an awesome resource for tunnelling

https://github.com/anderspitman/awesome-tunneling

I have played around with frp, bore and ngrok.

Gemini 3.8 flash has better rates. $0.75 per million input tokens and $3.75 per million output tokens.

Compare that to Muse spark 1.3

$1.25/M input, $4.25/M output (without data sharing) $0.10/M input, $0.20/M output (with data sharing)

It is dirt cheap, but only if you are willing to share your data with meta and allow them to use it for improving their models and products.

it seems like gemini 3.8 flash is more capable and cheaper. The only reason i would use this is if i was willing to share my data with meta, and allow them to train on my data. In that case it becomes dirt cheap.

performing glitch attacks on embedded systems, consoles, phones etc.

Maybe compatibility goes away a with TUIs. But portability, speed, remote access, and less resource usage are still key advantages.

TUIs, work on several operating systems. I use same set of apps across Mac, windows and Linux.

You can use them remotely over SSH connections, and they can survive disconects (within tmux). No VNC, or X11 forwarding needed.

They take less memory. compare vscode vs nvim.

Super fast to launch.

They can integrate with other TUIs and shell apps, shell-scripts, pipes

and lastly they do not contain Ads. lol

The script is blocked by our interal firewall. The reason : malware.

I tried using it on my x86_64 desktop. Could not use it.

Then I tried using it on a raspberry pi 4. Did not work.

This is specifically for 64 bit arm. I see no reason why you could not enable desktops as well? Just a matter of compilation

This is excellent model. I was working on some Linux kernel code, and Sonnet 5, Opus 4.8 had given up on the problem i was trying to fix (after several hours). Opus 5 was able to triage and fix the issue in under 30 minutes.

We symlink AGENTS.md and CLAUDE.md to a single file in our repo

California is one of those states that does not allow FSD usage to impact insurance rates

lol, even the link is not correct.

.con, instead of .com

Usually the first thing, the page should mention is how to install it.

it is as if developer does not want anyone to use this tool.

This government contract was a very small part of anthropic revenue. Almost negligible. Their 2026 revenue is projected to be $14 billion to $20 billion. This contract was $200M over 2 years.

it can be used as a novel way to host files.

Airplanes and many other captive portals will allow DNS traffic, but restrict everything else. Such things can be used to get free internet in such environments. It is indeed an abuse of protocol, and future protocols are going to make life difficult for everyone to prevent such abuse.

I already loved Claude models, and this makes me even more eager to use them.

Paramount agreed to pay the $2.8 billion breakup fee that WBD would owe Netflix if that deal didn’t go through.

I liked the simplistic UI, but it is not quite ready for serious use.

I tried using it on a simple svg that i had (around 1KB, just few simple lines and shapes). But it did not rendered them properly. Colors were off (black-box instead of original colors), and in one place it was showing at triangle instead of an L shaped line.

Also when I move an object, a single Cmd+Z wont undo the action. Have to repeat twice for object to go back (I am using Chrome on macbook)

Using claude code with custom models

Will it work? Yes. Will it produce same quality as Sonnet or Opus? No.

Open models are trained more generically to work with "Any" tool.

Closed models are specifically tuned with tools, that model provider wants them to work with (for example specific tools under claude code), and hence they perform better.

I think this will always be the case, unless someone tunes open models to work with the tools that their coding agent will use.

I was skeptical too, but after trying lazygit, lazydocker, k9s and a few more TUI programs, I can see the value.

NAT is the reason for IPV6 not taking over.

Also it acts as a nice security perimeter. If all IoT devices in a home were exposed to internet, It would be absolute mess.

1. replacing junior engineers, with AI ofcourse breaks the talent pipeline. Seniors will retire one day, who is going to replace them? Are we taking the bet, that we wont need any engineer at that time? sounds dangerous.

2. Junior engineer's heavy reliance on AI tools is a problem in itself. AI tools learn from existing code that is written by senior engineers. Too much use of AI by junior engineers will result in deterioration of engineering skills. It will eventually result in AI learning from AI generated code. This is true for most other content as well, as more and more content on internet is AI generated.