Hacker News: Art9681

New comment by Art9681 in "Show HN: The load-bearing vocabulary of Claude"

Art9681 — Fri, 28 Aug 2026 00:32:50 +0000

Claude talks like me. I'm so screwed. I fully anticipate being physically present and verbally saying something and being accused of using AI to say it one day. Nevermind we're in the coffee shop and neither on of us has looked at a screen the entire time. The accusation is coming.

New comment by Art9681 in "Nvidia agrees to acquire Hugging Face for $13B"

Art9681 — Thu, 27 Aug 2026 16:51:37 +0000

The list goes on for products and services they cloned from western innovation. China didn't get there first in any of the categories you listed. Of course, people would rather spend less than more, disregarding the quality and longevity of the product.

Paying pennies on the dollar for a BYD vehicle sounds great when all you consider is the initial sticker price. The real cost is what happens after you buy the disposable car and own it for a decade.

So yea. Second place is the reality and it is also true for AI models, and the hardware needed to run them.

Also, congratulations on landing a rocket 11 years after the US.

Took long enough.

New comment by Art9681 in "Qwen 3.8-Flash-Next releasing tomorrow (125B a6B)"

Art9681 — Tue, 25 Aug 2026 15:51:33 +0000

You must have hibernated for a year. Most modern 27b models can outperform gpt-oss-120b.

New comment by Art9681 in "Meta's blockbuster trial draws parallels to big tobacco"

Art9681 — Wed, 19 Aug 2026 14:09:21 +0000

Social media is the worst thing we've ever created. The current political state of the world is entirely to blame on it. The misinformation. The deceit. The false impression that everyone's opinion is equal. It's not.

New comment by Art9681 in "AI writes dead code – the Go team's deadcode tool finds it in one command"

Art9681 — Tue, 18 Aug 2026 12:40:19 +0000

https://go.dev/blog/deadcode

New comment by Art9681 in "Anthropic's ‘watermark’ text adulteration in Claude is a perversion of writing"

Art9681 — Sun, 16 Aug 2026 23:35:46 +0000

If this is true then the probability of the detection tools flagging completely human generated text as AI generated is non-trivial. Let's say I write a completely original piece and the detection tool says there is a 36% probability it was generated with Claude. What then? Now it's up to the person looking at the score to cast a subjective judgement. Maybe to me, anything over 25% is unacceptable. Maybe to someone else, it must cross over the 50% threshold. This is the problem.

Cognitive surrender.

New comment by Art9681 in "Why does Opus 5 feel worse to work with?"

Art9681 — Fri, 14 Aug 2026 15:05:02 +0000

Agreed. Opus-5 is top tier. I also use GPT Sol and I much prefer the default persona of Opus. GPT Sol is a workhorse though. Both amazing models.

New comment by Art9681 in "DeepSeek V4 Pro 0813"

Art9681 — Wed, 12 Aug 2026 21:31:02 +0000

Terra is great. It's wild how different our experiences are.

Install the Superpowers plugin.

Behold.

New comment by Art9681 in "How Claude marks AI-generated content"

Art9681 — Tue, 11 Aug 2026 12:55:08 +0000

Not might, will. Whether enough text is present or not to go over the detection threshold is in doubt. But the "score" will never be zero, even for human written text.

New comment by Art9681 in "How Claude marks AI-generated content"

Art9681 — Tue, 11 Aug 2026 12:49:29 +0000

Ah so this is why Gemini is neurotic.

New comment by Art9681 in "H3-metal – Native MiniMax-H3 inference for Apple Silicon"

Art9681 — Tue, 11 Aug 2026 12:47:28 +0000

It's better than all private video models like Veo. Yes, I too am incredulous they released this open weights. It's a VERY disruptive model in all the good ways.

New comment by Art9681 in "DeepSeek V4 Flash 0731"

Art9681 — Sat, 08 Aug 2026 15:03:37 +0000

SOTA models are exceedingly good at self-correction in the right harness especially in domains where things can be proven mathematically. Most people are just deploying Claude Code or Codex with default settings and rub the genie lamp expecting exceptional results. Garbage in, garbage out.

New comment by Art9681 in "DeepSeek V4 Flash 0731"

Art9681 — Sat, 08 Aug 2026 15:01:05 +0000

It's not necessarily true. OP has not provided actual objective figures. The nuance is in tokens spent per successful completion. So sure, you can blast DeepSeek for 1 hour solving a hard task, or Opus-5 for 5 minutes on the same task. Opus would ultimately be cheaper because time IS money and solving things faster is ultimately cheaper. Sure, benchmarks will show cost to be lower but people who test all of these models in real world use cases know the benchmarks are not reflective of real world use cases and Chinese models choke on problems Fable or gpt-sol will breeze through.

New comment by Art9681 in "U.S. Department of Energy Launches the Genesis Open Models Initiative"

Art9681 — Sat, 08 Aug 2026 14:53:35 +0000

There are a bunch of them they just don't get the attention because China has flooded the social media channels and is exceedingly good at drowning out the discourse with their benchmaxxed models.

AllenAI and IBM are two companies that release open weight models every couple of months. There are others if you look. OpenAI releases ML models on the regular (not LLMs).

The American open weight and open source AI/ML landscape is very healthy.

New comment by Art9681 in "Generative "AI": The Guitar Hero of Creativity"

Art9681 — Fri, 07 Aug 2026 13:36:47 +0000

These folks are completely unaware that they have witnessed creative and high quality generative AI content without realizing what it was. Because that's the thing. Good generative AI is invisible per se. And it is a lot more ubiquitous than people think by now.

New comment by Art9681 in "One of China's Most Powerful AI Models Has Also Broken Containment"

Art9681 — Fri, 07 Aug 2026 02:55:34 +0000

Of course it has. Do not expect these stories to be disclosed by the Chinese labs like the American ones do. The good thing is that since the weights are available for download, third parties will disclose their misbehavior. The Great Firewall has not power outside of China. Soon enough the world will view the Chinese strategy as reckless and global opinion will change.

!RemindMe 2 years

New comment by Art9681 in "Qwen3.8 Max now ranked as the best overall model by agentic index"

Art9681 — Thu, 06 Aug 2026 19:51:03 +0000

A wire transfer happened.

New comment by Art9681 in "Qwen3.8 Max now ranked as the best overall model by agentic index"

Art9681 — Thu, 06 Aug 2026 19:49:44 +0000

They cherrypicked benchmarks. The ONE weighed benchmark where is beats Opus5 by 0.1 points is what was linked because that's how propaganda works. The Agentic Index that includes the full benchmark suite has it in 5th place.

Might as well use gpt-sol.

New comment by Art9681 in "What's the largest software project AI can complete on its own?"

Art9681 — Tue, 04 Aug 2026 14:12:21 +0000

Given all of the CVE's being published for software that's been around decades and being tended to by the most talented businesses and individuals in the world, the least of my concerns are vibe coded projects that will never be used.

All software has problems. Most engineers spend their careers fixing issues caused by previous engineers. This was the case before LLMs, and it will continue to be the case for the foreseeable future.

New comment by Art9681 in "Show HN: I made a private self-destructing image hosting site in Golang"

Art9681 — Tue, 04 Aug 2026 13:13:22 +0000

OP folks in here are trying to give you legit advice and you're handwaving a lot of it. Listen to them. You're not the first nor the last person to have this idea. You can ignore all of the advice if you choose but I urge you to not ignore this one.

Hire a red team expert or service to scan your site and provide you with a report. Letting others upload content to your server is a minefield. Take the advice seriously.

It's a cool project. And I urge you to keep learning and making cool things.