Hacker News: anon373839

New comment by anon373839 in "California is chasing wealth that has feet"

anon373839 — Thu, 24 Sep 2026 23:26:41 +0000

> the argument that billionaires will leave if taxed at a higher rate isn't compelling

I agree with this. California’s climate and culture will keep many a billionaire within tax nexus reach of the state. Of course, this isn’t a strategy every locale can pursue, but I don’t see a reason for California not to exploit its advantages.

New comment by anon373839 in "California is chasing wealth that has feet"

anon373839 — Thu, 24 Sep 2026 23:22:34 +0000

> Rent is a function of supply and demand

Is there really a market dynamic in rent pricing anymore? I thought that algorithmic collusion had eliminated the need for landlords to compete on price.

New comment by anon373839 in "Claude Opus 5.5"

anon373839 — Wed, 23 Sep 2026 06:24:02 +0000

The fix for this is to tap the overflow menu icon and choose “Reduce privacy protections”. (Wtf, Alibaba?) This appears to be related to use of iCloud Private Relay.

New comment by anon373839 in "MiMo v2.6"

anon373839 — Tue, 22 Sep 2026 07:37:30 +0000

The gap really is closing, though. I don't remember the source, but there was a publication recently stating that it's 4 months at this point. So, 4 months where you can charge premium prices for the small subset of tasks where only a frontier model will do. I have to think the big AI duopoly is at an inflection point where most of their customers haven't just yet realized that they're being fleeced.

New comment by anon373839 in "Fable 5 – Median thinking declined in August"

anon373839 — Tue, 22 Sep 2026 02:50:14 +0000

Qwen 3.8 Flash-Next is not dumb. If you've used it and that was your experience, your workload is either ultra-ultra-sophisticated or you're dealing with a broken quant/buggy chat template/other issue. That model is a smart, reliable workhorse.

New comment by anon373839 in "Alibaba open-sources AI model that can detect cancer and nearly 150 conditions"

anon373839 — Sat, 19 Sep 2026 04:42:26 +0000

The model weights are only ~5GB, so this is small.

New comment by anon373839 in "Claude Code now reads AGENTS.md if there is no Claude.md"

anon373839 — Fri, 18 Sep 2026 23:59:40 +0000

The world has moved on.

New comment by anon373839 in "Qwen 3.8 Omni Flash"

anon373839 — Fri, 18 Sep 2026 05:41:25 +0000

This is a serving bug or quantization issue. I had all kinds of issues that were like this on DGX Spark until I found a single-GB10 vLLM recipe [1] that uses Nvidia's NVFP4 quant. The community quants did not work well.

Another failure mode you may see is inordinately long CoT. Properly served, the model is good at calibrating its CoT length to the difficulty of the immediate task.

[1] https://github.com/blazux/qwen3.8-Flash-DGX

GVS5H: Five Qwen3.8 Models Match Claude Fable 5 on LiveCodeBench Hard

anon373839 — Thu, 17 Sep 2026 13:46:04 +0000

Article URL: https://github.com/slee-persis/GVS5H

Comments URL: https://news.ycombinator.com/item?id=49740680

Points: 2

# Comments: 0

New comment by anon373839 in "HarnessTax: How Much Does the Harness Matter for Coding Agents?"

anon373839 — Thu, 17 Sep 2026 09:27:01 +0000

> Personally, I’ve replaced OpenCode with a thin wrapper around Pydantic-AI as the pythonic analogue to Pi-Agent for headless use via Hermes

That's really interesting. I like Pydantic AI a lot and wondered why all of the harnesses seem to be written in Javascript instead of it. What do you use it for headless, though? I haven't tried Hermes or similar yet, so don't have a handle on what you do with them.

New comment by anon373839 in "A warning about 'model welfare'"

anon373839 — Wed, 16 Sep 2026 23:19:59 +0000

Model welfare, much like AI xrisk, is a concept born from evidence-free “what if?” questions. Some people ran with these what-ifs and developed ornate belief systems around them. And now they demand the rest of us take them seriously.

New comment by anon373839 in "Due to concerns about malicious applications, GPT2 will not be released (2019)"

anon373839 — Mon, 14 Sep 2026 00:48:23 +0000

That isn’t any kind of antitrust violation. Agreeing not to undercut each other’s prices would be, however.

New comment by anon373839 in "Everyone should slow down AI development except for me"

anon373839 — Sun, 13 Sep 2026 23:58:05 +0000

Ah, no, that’s not cheaper. Renting GPUs adds up quickly and leaves you with nothing in the end.

Renting tokens from open model providers is cheaper but it incurs the same issues: unexpected changes in model quality, inconsistent speeds, service outages.

New comment by anon373839 in "AI models don't kill people – people kill people"

anon373839 — Sun, 13 Sep 2026 14:38:21 +0000

I will say that LLMs are somewhat unlike guns in that they emit text.

New comment by anon373839 in "Everyone should slow down AI development except for me"

anon373839 — Sun, 13 Sep 2026 13:55:53 +0000

Privacy is a great reason, but independence is another. It’s very nice knowing that you’re going to get the same reliable product every time you call the model. Nothing is going to change unless you decide to change it.

New comment by anon373839 in "We must pace the frontier"

anon373839 — Sun, 13 Sep 2026 13:52:24 +0000

It’s absolutely laughable that he refers to METR as if they were neutral observers. They are ex-Anthropic employees and others with direct financial interests in Anthropic.

New comment by anon373839 in "Everyone should slow down AI development except for me"

anon373839 — Sun, 13 Sep 2026 06:12:48 +0000

It is costly, especially right now. I don’t think you can make a case for it on cost savings!

The throughput in a single stream is about 50 tokens/sec (a bit less for prose, a bit more for code due to speculative draft acceptance rates) and about 2,000 tokens/sec for prefill. Both numbers are flat and stable as context accumulates. That’s what finally tilted me away from the Mac Studio despite its much superior memory bandwidth.

I think these numbers may improve because the model is pretty new and optimizations aren’t done.

New comment by anon373839 in "Everyone should slow down AI development except for me"

anon373839 — Sun, 13 Sep 2026 05:12:20 +0000

> If someone put in frontier AI models from like .... last june I guess? in a box and let me run it with "decent" token throughput I would be happy.

You can have that! Qwen 3.8 Flash-Next is ~Opus 4.6 and runs nicely on a DGX Spark. And that’s just an architecture preview. The Qwen 4 family is expected to arrive this fall.

New comment by anon373839 in "Nvidia is the central bank of AI"

anon373839 — Sat, 12 Sep 2026 23:06:32 +0000

No, it’s not the active parameters. Qwen 3.8 Flash has 6B active and it smokes both models.

New comment by anon373839 in "ChatGPT: "Super shady" data controls settings"

anon373839 — Wed, 09 Sep 2026 12:15:31 +0000

https://xcancel.com/edoardocontente/status/20976073405059281...