likium

Will the model even be competitive in 10 months though? Seems like models that reach top 20 on OpenRouter see 50% of all token spend by day 80, and 80% by day 180.

Don’t worry. By the time they grow up they’ll just ask the AI for a summary of their childhood.

Here’s Anthropic’s announcement post: https://www.anthropic.com/news/confidential-draft-s1-sec

My guess is Gemini Pro coming later will be 2x more, bringing it comparable to Opus’s pricing.

"Parable of the broken window", except instead of preventing the kid from throwing the rocks, they hired a someone to catch the rocks midair.

We didn't see the majority switch from Google to Duckduckgo because of ads or privacy... Being the "default" brings network effects that is hard to switch away.

Football anime doesn’t involve real people or stakes. AI can introduce a storyline, characters, etc. It won't necessarily be as popular as the real sport but I doubt the audience is zero.

For TypeVar I’d reach for a lint warning instead.

As a solo dev who picked SolidJS, yes, it is a big factor. Things like tldraw, node editors, wysiwyg editors, etc. Having to reimplement them is a huge time sink.

There's 3x more React libraries and code out there to reference. AI agents do _a lot_ better with React than, say, SolidJS. So productivity > performance is a tradeoff that many companies seem to happily take.

At least for the real life pictures, there’s no depth of field. Everything is crystal clear like it’s composited.

For $10 flat per request up to 128k tokens they’re losing money. 100 * 100k is 10m tokens. At current api pricing that’s $50 input tokens, not even accounting for output!

Just the decision of whether to allow models to invoke them has [1][2][3] different ways.

[1]: https://code.claude.com/docs/en/skills#control-who-invokes-a... [2]: https://opencode.ai/docs/skills/#disable-the-skill-tool [3]: https://developers.openai.com/codex/skills/#enable-or-disabl...

HN crowd did not like AI coding until it got better. How much of the AI hate is because of poor implementation?

So set up e2e tests and make sure it does things you said you wanted. Just like how you use a library or database. Trust, but verify. Only if it breaks do you have to peak under the covers.

Sadly people do not care about redundant and verbose code. If that was a concern, we wouldn't have 100+mb of apps, nor 5mb web app bundles. Multibillion b2b apps shipping a 10mb json file just for searching emojis and no one blinks an eye.

Equipment for any real life hobby costs money, and people are willing to pay for good equipment. Why is software so different? FOSS has made us so resistant to paying for good software. I'm not saying software has to be expensive, even $20 per major version could make a burgeoning indie industry thrive.

Supposedly Google made their own results worse to improve ad revenue. And I don't see mass migration over to Kagi or Bing.

Companies will pay OpenAI to prioritize more of their content during training. The weights for the product category will now be nudged more towards your product. Gartner Magic Quadrant for all businesses!

Go plans at $8 are getting ads too. Netflix introduced a paid plan with ads, and it is more profitable.

https://lik.ai - Recently been adding tools like color convertors and color contrast visualizers.

If someone painstakingly hand sculpted an exact replica of "David", does it make it art, or a forgery? Is hand written code to produce generative art not art?

It's difficult to pin down the line. Ultimately it's up to the individual to define them. "The relationship to art, and this kind of painting, to their work, varied with the person entirely."[1]

[1]: https://news.berkeley.edu/2025/03/31/berkeley-voices-transfo...

Tailwind requires a compiler to work.

It really depends on whether coding agents is closer to "compiler" or not. Very few amongst us verify assembly code. If the program runs and does the thing, we just assume it did the right thing.

The tool explicitly has an allow list for commands and block list for sensitive paths[1].

Though it's not enough, stuff like `grep "" .e?v` can still end up sending your sensitive keys to LLM providers.

[1]: https://github.com/ygwyg/system/blob/b5adfe526da7470cade61f7...

I do like the standalone HTML file approach. With coding agents it's quite easy to build a suite of tools for personal use. Additionally if you review the code, you can trust if it's really private/secure.

The ones on mine are more visual focused since cli tools are better at conversion, formatting and such.

It doesn't help that you can do a lot more indoors now, and the indoors has gotten more addictive. So relative to that, there's even less you can do outdoors.

Some PM: users are locked in to our ecosystem; they can't load other apps after ours! /s

But for real, the average number of apps people downloading get fewer year over year. When the most popular/essential apps take up more RAM, this effect will only exacerbate. RAM prices have also doubled over the last 3 months and I expect this to hold true for a couple more years.

Nowadays platforms seem to track IP addresses and other signals to grant limits. Using a VPN works.

Unfortunately, most people just want eggs, not the burden of actually owning the goose.

Citing is not referencing links, it's finding primary sources and fact-checking. Kurzgesagt[1], an informational YouTube channel, has had issues with LLMs citing LLM generated content.

[1]: https://youtube.com/watch?v=_zfN9wnPvU0&t=175