jokethrowaway

Once companies start realizing how much they're spending in AI to OpenAI and Anthropic for not much more employee output (the bottleneck is always the initiatives, not the code), they will look for cheaper options and they will eventually discover the chinese models.

My AI pilled clients who were early AI adopters are already there and they are looking for solutions to spend less.

Qwen3.8-Next, thanks to its new architecture, is quite fast even if part of it is streaming from disk

this feels around 1000 elo

crazy to see 110mb.com links in 2026 (on the github repo project), it's been a while

The US empire is ending.

Probably when the Chinese burst the AI bubble the whole fake web economy will come crashing down and close this 30 years of inflated money.

If you want an expensive model to reason on your files, you need to give them your files.

If you think a cheap model is smart enough to filter information to give to your expensive model, you can save some money. If you think your cheap model is smart enough to format your expensive output, you can save some money.

In practice, this didn't work well until Qwen 3.8.

Qwen 3.6 and (abliterated) Gemma 4 were almost there but still making mistakes.

They have cool tech but they can't do product, like most of big tech.

Have you tried wrangler? holy

Hire enough product owners and that's what happens.

That's why they need to buy startups every once and then, to bring some good bacteria in their messed up corporate gut.

The obvious goal is to destabilize the western economy and prove that US tech is a worthless bubble - but I agree, OSS AI is great for everybody and what OpenAI was supposed to be

I've used their gemma 4 quants since when they were not still working in llama.cpp and ik-llama.cpp and I don't remember any problems

They are the most reliable in my experience, but if you have alternatives you trust I'd love to know

Very cool and I wish you the best of luck (I genuinely hate this generation browsers) - but my experience with bun and other complex zig tools is that a segfault always awaits behind the corner

likely just opus being dumbed down to prepare for fable

Great work! Always love more competition driving better products.

I'd make a non competitive mode without mistakes system: I love the puzzles, idgaf about leaderboards

It is not.

Eventually a new startup will replace your large inefficient employer with people working 10% of their time.

You didn't prove him wrong.

You and Paul are both claiming it's extremely hard.

A restaurant won't scale - most ideas won't scale - because they're not useful enough.

These companies are not criminal enterprise and they don't steal money or threaten them with violence: they just offer them products and ecosystem and most people are dumb enough to ruin their life with it. Don't get me wrong: I hate them and I ban social media for my kids, but it doesn't make them evil.

The bad actors are thugs with guns stealing your purse and the government stealing your tax money or threatening you with jail.

To give another, less controversial example, I don't think companies selling products with sugar and seed oils are bad, even though they likely have the highest combined reduction of life expectancy across all people - but I hate them too.

I never understood the appeal for cosmopolitan.

I'd rather have platform specific minimal binaries than a single binary with hacks.

Installing packages is a solved problem

Amazing stuff!

I would recommend an add-on or new feature to detect and remove cookie banners / annoying popups that open on load (eg. sign up to my mailing list).

listing a few examples form fastText could help you.

You might also have the opposite problem though: some websites have content in the base html (so it's searchable by Google and they get views) and remove it on load (so you have to pay).

Capturing the initial html and comparing it to the final version could give you some hints and allow you to repair the removed content.

Best of luck with the project!