apitman

Would love to have this on https://github.com/anderspitman/awesome-tunneling as soon as it hits 100 stars.

Wow this thing just blew out the pareto front for intelligence/$

See Cerebras and Groq as well.

Speed is the metric I'm currently most interested in. The models are smart enough.

Once speed significantly increases I think we're going to see some interesting downstream effects. The three things I currently spend the most time waiting on are LLM API requests, Rust compile times, and nix derivations. As AI latency approaches zero I think we're going to start taking a hard look at whether slow-compiling languages are adding enough value over Golang, Typescript, or even dynamic languages to be worth the slowdown.

That's surprising. Does it have no sub agent support at all or does it just use the same agent as the parent?

> The way to solve for this is to have them work on two different worktrees with separate environments, ports

What are you using for the environment part? Containers?

> More importantly, I don’t enjoy how people interact on social media anymore

I'm really just not convinced social media is very useful. The main use I can see is very fast dispersal of information, but anything truly important will be distributed by mainstream media, and less important things will travel between small communities at a slower rate. So what are we left with exactly?

I think small communities where people actually have to maintain a reputation are the future, and the past.

I got hit with my first belt-and-suspenders by Kimi K3 this morning. I normally use GPT. Is that a Claude-ism?

Ah gotcha, I hadn't noticed that. Thanks

Are you sure the difference is from MoE and not that 3.6 is newer?

Artificial Analysis ranks qwen3.6-27b higher than qwen3.5-122b-a10b on both intelligence and coding. Does that run counter to your experience?

Honestly it's pretty wild that the standard way to download these isn't torrent instead of Hugging Face direct. Why doesn't HF themselves provide torrent links?

This is my experience as well. The two areas I get the most leverage out of AI are root cause analysis and vibe coding green field prototypes.

Sunshine was significantly easier for me to get working using a dummy plug. I'm currently exploring other options but that's definitely the happy path.

FWIW Incus is really good here. I don't quite have it fully working for headless gaming containers sharing a single GPU, but I'm very confident it will work at this point.

The line separating HN and reddit runs through the heart of every man

The bones are their money

In my utopia traditional advertising doesn't exist. Products are in catalogs. When I want to buy something I look in the catalog under a specific category. Innovative products that actually bring value are spread by word of mouth.

I run my agents in a VM. They don't even have access to their own LLM API keys.

The 5 hour limits are coming back soon right? I thought that was temporary

> give me Opus 4.5 at a tenth the cost and running ten times as fast and I'd take that for almost any coding task over any current frontier model

This is pretty much where I'm at

Just wanted to say I started using Handy last week and I love it. It might single handedly cure my RSI. Well, hopefully double handedly.

Drug store cowboys, aka "big hat; no cattle"

I can only imagine what that does to the poor SSD

What kind of tokens per second do you get on that setup?

Have you tried qwen3.6-27b or Gemma 4? gemma4-26b-a4b is particularly impressive for its balance of speed and performance

> Helping people write word docs, recipes, and emails isn't going to justify $15K per month subscriptions either.

Those things can all be done today on a $250 used video card and pennies of electricity

It would be more efficient if you had multiple users (or agents) making parallel requests to take advantage of batching, right?

I've got dual 3060s as well. What's the best models you've found for this setup?

You could test this by starting your sessions with "I am Terrance Tao"