Hacker News: efficax

New comment by efficax in "GLM-5.3-Flash"

efficax — Wed, 26 Aug 2026 23:12:26 +0000

you don't have to hand them your data, the models are available so you can run them on bedrock yourself (or use another US housed inference service). and for what it's worth in my job i have access to data that gives a picture of the way companies are doing inference, and they're using a lot of chinese models (deepseek-v4 is a huge percentage of inference requests for example)

New comment by efficax in "Anthropic expected to tell investors it sees over $30T in potential revenue"

efficax — Tue, 25 Aug 2026 19:28:09 +0000

Reuters should know better. While $30T is probably absurd, Anthropic is claiming that's the total addressable market for AI services, not that they will capture 100% of that market

New comment by efficax in "New Mac mini, featuring M6 and M5 Pro"

efficax — Tue, 25 Aug 2026 19:19:32 +0000

yeah my m1 mac studio shows no signs of stopping and i can't justify another upgrade until doing so really unlocks something new at a reasonable price (like running local llms that are competitive with the cloud services at a decent speed, which presumably the mac studio M5 Ultras can do with qwen 3.8 but the 256GB Ultra config is more than $10k!)

New comment by efficax in "New Mac mini, featuring M6 and M5 Pro"

efficax — Tue, 25 Aug 2026 14:26:00 +0000

they've done a bunch to make macs a better gaming platform, including the very nice game porting toolkit (https://developer.apple.com/games/game-porting-toolkit/) which you can use yourself to run pc games on the mac. All macs since m1 have very nice gpus that can handle most games made today, and the newer ones can run the latest AAA ones with good detail. At this point it's more that game publishers/studios don't care about the mac

New comment by efficax in "Coding expertise is going to collapse from AI reliance"

efficax — Mon, 24 Aug 2026 18:49:29 +0000

And so what? How do you think people are going to be working on software in 10 years. They're going to use coding agents and harnesses that work much better than the current ones (although I don't think we're getting anything but incremental improvement from here on out). They will run on commodity hardware. I first got paid for writing code in 1999, I've worked at publicly traded software companies and failed startups, in every language out there. And if LLMs "atrophy" the expertise I built up over those years, then so what? What is lost, exactly? I stopped programming in C++ about 10 years ago when Rust seemed to get good. I no longer remember a lot of the specialized knowledge I had in C++. My life goes on. I am now becoming an expert at building software with agents. The problem is still the same, you still have to apply engineering principles. You just don't have to write the code.

New comment by efficax in "Quick impressions: A week of using Codex more than Claude"

efficax — Fri, 21 Aug 2026 23:24:34 +0000

you hear things like this about every model. they're probabilistic and you can't trust them, although i doubt that kind of odd behavior would happen on low or medium effort levels. i find all of the models today pretty good at following instructions generally speaking. i'm sure you did /plan, but i rarely see it stray very far from the plan. Then again, I never walk away, since i'm always spinning up another task that can be done in parallel, and I don't let things run overnight because I'm not made of token cash (yet)

New comment by efficax in "DeepSeek-v4-flash-vision-exp"

efficax — Fri, 21 Aug 2026 21:32:20 +0000

fwiw the clock reads 8:09:25 not 8:10:25 although it would be passable to read it as 8:10 if there were no second hand

New comment by efficax in "Copyright does not protect AI-generated content in EU"

efficax — Fri, 21 Aug 2026 00:52:14 +0000

Does this apply to software?

New comment by efficax in "Incident with Github.com [resolved]"

efficax — Mon, 17 Aug 2026 15:12:38 +0000

I think AI is going to change this. Once you've built around github you feel locked in, since moving all your CI and other actions out of there, and your workflows out of there, is labor intensive. Now you could migrate from github to another service in a week, maybe less depending on how many workflows you need to move.

New comment by efficax in "Show HN: Pixy, visual editor for coding agents, like Figma on your live site"

efficax — Thu, 13 Aug 2026 19:31:36 +0000

Interesting idea, not at all clear to me what it has to do with coding agents

New comment by efficax in "Qwen3.8-2.4T"

efficax — Wed, 12 Aug 2026 18:43:17 +0000

they explicitly say this is qwen3.8 max in the hugging face page

New comment by efficax in "OpenAI’s head of ethics leaves less than a year after joining"

efficax — Tue, 11 Aug 2026 21:54:05 +0000

staying for less than a year usually means you don't take any equity with you, it must've been an incredible offer in that case, since those openai shares are a golden ticket right now

New comment by efficax in "Mojo 1.0"

efficax — Tue, 11 Aug 2026 21:08:13 +0000

i think the main idea is that by using MLIR[1] instead of the usual LLVM path it can target gpus directly

1. https://mlir.llvm.org/

New comment by efficax in "Closing Canario Terminal source code"

efficax — Tue, 11 Aug 2026 11:50:17 +0000

Do what you want, sure. But people complain a lot about "open source maintenance". But you could just release the source and call it a day. Put up a web page with a tarball of the code, updated whenever you consider it a new release. You don't have to publish your git repo and allow pull requests and issues and all of that nonsense. None of that has anything to do with the purpose of open source, which is about giving access to the code that runs your computer so you can be in control of it.

New comment by efficax in "Humanising LLM Outputs Is Dumb"

efficax — Mon, 10 Aug 2026 20:20:46 +0000

is it lossy though? That didn't make sense to me. You can tell it to use Simplified Technical Language and also still have it give you all the detail. it's just another piece of the prompt that produces the output. it's not like there's "pure" llm output and then "lossy" output guided by a prompt.

New comment by efficax in "Squeak 6.1"

efficax — Mon, 10 Aug 2026 16:08:52 +0000

Glamorous Toolkit is built on Pharo, which I think is a fork of Squeak or shares lineage with it.

New comment by efficax in "You might want to build your WebApp in Canvas instead of HTML"

efficax — Sun, 09 Aug 2026 11:14:12 +0000

Not a word in here about accessibility, of course. If you’re blind you’re blind to the canvas as well without a lot of work, it’s much more straightforward to make the DOM accessible

New comment by efficax in "Melatonin impairs morning cognition in healthy young adults (2023)"

efficax — Sun, 09 Aug 2026 11:06:50 +0000

at least!

New comment by efficax in "“Code was never the hard part” is an insult to all programmers"

efficax — Sun, 09 Aug 2026 00:26:37 +0000

writing code is hard. building a successful product is harder, and is only sometimes a consequence of the code.

New comment by efficax in "Timeline of the OpenAI accidental attack against Hugging Face"

efficax — Sat, 08 Aug 2026 14:33:23 +0000

well sure, but i mean realistically speaking, we cannot step debug an llm's output to find out what happened given the way we currently execute inference