Qwen3.8-Max: A New Bar for Coding and Cowork • Hacker News Board

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#621

Post by CyberDildonics » Tue, Aug 04, 2026, 12:14 PM UTC

lcnPylGDnU4H9OF wrote: post #607

CyberDildonics wrote: post #595

Earlier quoted context omitted.

Saying, well a rule of thumb is, well, 100 billion bytes is a 100 gigabytes, is well, not a rule of thumb. It is, well, just the common definition.

Right. The rule of thumb is that the overhead size of the model that's not the weights is so vastly outweighed by the actual number of weights that it can be disregarded. My shorthand for that was to write "the weights take up ~100% of the size of the model". What then "follows", both in the sense that the explanation is written after the rule as well as that it logically follows, is that, well, 100 billion bytes is,…

Right. That's, well, not a rule of thumb. Asking how much of a bottle of water is, well, water, and someone says "well, a good rule of thumb is that it's all water", is just an answer. You don't need an estimate when, well, there is nothing to estimate.

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#622

Post by jiggawatts » Tue, Aug 04, 2026, 12:17 PM UTC

ahartmetz wrote: post #620

jiggawatts wrote: post #578

Earlier quoted context omitted.

Three hours is a lot longer than one minute.

No shit, but the huge ordeal you described is an exaggeration.

"X is less than Y".

"X is literally more than Y".

"You are exaggerating how much X is!"

"It's still a lot more than Y."

PS: This whole thread reminded me of several managers I've worked with who were pathologically unable to estimate... anything, be it driving time or development effort.

They always focused on the "minimal aspect", ignoring everything before and after. Walking to the car park. Standing in line at the machine. Paying at the machine. Getting out of the car park in the car, surprisingly long during busy times. Driving through traffic. Any delays that could -- and regularly do -- occur. Finding parking. Actually parking. Walking from the car park. Etc.

"It's just a 5 minute drive!"

No, it isn't, not door to door.

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#623

Post by ahartmetz » Tue, Aug 04, 2026, 12:23 PM UTC

jiggawatts wrote: post #622

ahartmetz wrote: post #620

Earlier quoted context omitted.

No shit, but the huge ordeal you described is an exaggeration.

"X is less than Y". "X is literally more than Y". "You are exaggerating how much X is!" "It's still a lot more than Y." PS: This whole thread reminded me of several managers I've worked with who were pathologically unable to estimate... anything, be it driving time or development effort. They always focused on the "minimal aspect", ignoring everything before and after. Walking to the car park. Standing in line at the…

In which case, are you sure about your one minute?

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#624

Post by Alifatisk » Tue, Aug 04, 2026, 12:35 PM UTC

wei_b0 wrote: post #610

Alifatisk wrote: post #152

Earlier quoted context omitted.

Already done it. But switched to Codex instead. See you on the other side buddy.

I use codex as well but the limits have become absurd though - end up spending weekly budget in like 2 max 3 days

I’ve started to offload some tasks to Luna since the dramatic price change. If you crank up the reasoning effort to High or Xhigh, it is quite good! I previously used Sol (medium). I suggest to give it a try, you also save so much on usage.

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#625

Post by lcnPylGDnU4H9OF » Tue, Aug 04, 2026, 12:54 PM UTC

CyberDildonics wrote: post #621

lcnPylGDnU4H9OF wrote: post #607

Earlier quoted context omitted.

Right. The rule of thumb is that the overhead size of the model that's not the weights is so vastly outweighed by the actual number of weights that it can be disregarded. My shorthand for that was to write "the weights take up ~100% of the size of the model". What then "follows", both in the sense that the explanation is written after the rule as well as that it logically follows, is that, well, 100 billion bytes is,…

Right. That's, well, not a rule of thumb. Asking how much of a bottle of water is, well, water, and someone says "well, a good rule of thumb is that it's all water", is just an answer. You don't need an estimate when, well, there is nothing to estimate.

Yeah, well, that's just, like, your opinion, man.

Anyway, more seriously, I hope it's obvious by now that I don't particularly care that my means of communication is so offensive to you. I think you should, like, cry a river, build a bridge, then, well... get over it, you know?

And on the, er, "topic"? A rule of thumb for me does not have to be one for you, even if it's explicitly presented as such a rule. I thought that would be obvious but, well, here we are. Anyway, how's life been treating you?

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#626

Post by HarHarVeryFunny » Tue, Aug 04, 2026, 1:53 PM UTC

nerdalytics wrote: post #601

HarHarVeryFunny wrote: post #480

Earlier quoted context omitted.

Where did you read that Anthropic are experimenting with rewriting Claude Code?

It's a rewrite of Claude Desktop from Electron to SwiftUI as Theo 3G said in one of his last videos.

Thanks

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#627

Post by zeratax » Tue, Aug 04, 2026, 2:15 PM UTC

rrix2 wrote: post #48

mlmonkey wrote: post #7

Earlier quoted context omitted.

That's the thing. Wny are companies like OpenAI/Anthropic/Alibaba/Kimi/Deepseek still hiring SWEs if their models have become so good?

https://en.wikipedia.org/wiki/Jevons_paradox

i dont see how that answers the question? why not answer more demand with even more ai agents?

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#628

Post by Forgeties79 » Tue, Aug 04, 2026, 3:14 PM UTC

jiggawatts wrote: post #579

Forgeties79 wrote: post #458

Earlier quoted context omitted.

> Local: Typical scenario is hours just to download the software, the model weights, and then faffing around with CUDA and matching your GPU drivers. Download LM studio, search models, click download, wait minutes, prompt and have fun

“If you have the prerequisite hardware, then… know which model you want out of thousands of a variants… and your drivers are up to date, then it is fast!”

> If you have the prerequisite hardware

Literally tens of millions of people of silicon MacBooks have sold send 2020 so it’s probably safe to say hundreds of millions of people have the necessary hardware. Not even getting into smartphones.

>which model you want

Have you personally searched for models in LM studio? It’s actually pretty straight forward and it tells you with a very clear icon if it will all fit in your GPU or if it will offload onto ram.

>drivers are up to date

Are you just making things up now? I run LM studio on an M1 MBpro (albeit very small model for small tasks with tool calls) and on a Linux (fedora) PC with an AMD GPU. In both cases i downloaded LM studio, quickly found models with their search, and started messing around. I am not a coder or engineer mind you, so clearly it isn’t that difficult.

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#629

Post by ilaksh » Tue, Aug 04, 2026, 3:23 PM UTC

DANmode wrote: post #605

ilaksh wrote: post #548

Earlier quoted context omitted.

Can you give me a hint about wtf you are talking about?

Use AI to speed up your own existing workflows that were working. Advertise that you’re there to fix AI slop attempts (you can rebuild it from scratch if you have to). You may not have the juice to sell great shovels in a gold rush. You also may . Just…be sure.

I've been using my agent harness to research and build AI services for many months.

I'm honestly not that interested in cleanup work because most of what I see is not a genuine failure but just a vague request to "make it production ready" and half of them have the expectation that I will use little to no AI to do that and they think there is some magic set of best practices that will prevent any issues from popping up.. and they have quite low budgets.

Which is the type of busywork that less capable engineers overcharge for. What they need is usually modest cleanup and then a strong iteration loop. Anyway, it's not that easy to find projects that make sense.

Re: Qwen3.8-Max: A New Bar for Coding and Cowork

#630

Post by SummSolutions » Tue, Aug 04, 2026, 3:51 PM UTC

AbsurdCensor wrote: post #518

SummSolutions wrote: post #481

Earlier quoted context omitted.

Nice. Mind sharing the solar side of your setup?

Couple of rack mount batteries and roughly 5kw of solar panels. Feeds into a subpanel so I can flip it when I want a couple rooms of solar on the house, or hook a generator up if needed. Can't power the entire house, but works well for thinks like computers, lighting, etc. And if I want to expand, just throw on more panels, or realistically, just throw on more batteries to store the juice.

Thanks - that’s what I want to do!