Genuine question: can these posts be paid to hype the open source models? If yes, what would be the purpose?

On my work tasks, FastAPI Python and Springboot Java on a modern SaaS product, the only open model that can do tasks well and efficiently is Qwen3.7-Max.

In all my experiments, both GLM-5.2 and Kimi are busy grepping around the codebase for ALMOST 70-80K tokens before writing anything and when they do it typically breaks the code… it feels to me that these models are good but only when you write out a super detailed spec of the task just like it was done a year ago… Qwen3.7 just… does it

It's (good) content marketing. They sell access to Kimi K3. They're one of the biggest model inference providers out there.

Fireworks is an inference provider that specializes in running open source models very fast and makes almost all of their margin on chinese models

So the economic incentive is literally their entire business model lol

Lots of money in tech influencing - but most is from the big players (OpenAI buying tbpn, early access to select influencers etc).

>the only open model that can do tasks well and efficiently is Qwen3.7-Max

Qwen3.7-Max is a proprietary, closed weight model.

Have you tested Qwen3.8-Max yet?

Qwen is so much better than GLM or Kimi that this makes me genuinely excited!!

The bigger models are usually worse though, hopefully they nail it this time.

> Take a task, any medium-sized task, decently scoped that you'd trust to give to Sonnet to finish without a hitch. Now give it to ANY open-source frontier model and watch them struggle and go in circles while failing tool calls and randomly assuming things.

Claude used to be much worse than it is now, just as bad the open weights models are. And the open weights were worse. The labs will also try to keep the lead, but at some point people start seeing real value from open models. Maybe you say they're not ready yet for medium tasks, but everyone sees the writing on the wall.

I hope you're right and I want you to be right, but, even seeing the current hype around local models, etc... and open-source models, I think the industry is currently under a big confusion where they see the benchmarks of things like Kimi, GLM, Qwen, they play with it via opencode, and they think like: "Wow this is pretty good, I want to deploy this". But they don't understand how the KV cache grows over time and can take almost as much memory as needed for a 30B param model, they dont understand that a quantized model WILL NOT be the same as a full precision one, and they surely don't see the engineering work needed to serve inference to even tens of customers at a decent quality and latency level.

The biggest moat of these giant labs and models is increasingly shifting towards deployment capabilities and (debatably) having better (proprietary) harnesses.

The models themselves can be impressive on benchmarks, but unless they can be served reliably to customers either at scale, hosted somewhere, or even on edge with predictable latency and memory usage, then frontier will always be leading.

It sounds like you're focusing on the problems of running local models, or running models yourself, but I don't think that many people seriously expect near term improvement on that, it's definitely more just hopeful thinking there. That's not what I meant to address, and I also am in more of a "wait and see" mode.

But at this point we do expect that open weights _hosted_ options become feasible for the tasks they're using the frontier models for. And because of the lack of "legal monopoly" (intellectual property of whatever kind), they're way cheaper, not mention more flexible.

The launch of the tinker platform from Thinking Machines is an example of the "more flexibility" part that people want (and they chose to make their model open weights, maybe because this is the angle they want to push).

At this point I think it's realistic enough that the ball is in OpenAI / Anthropic's court to figure out how to respond to this threat to their business model.

That said, I think it's concerning that there are apparently only a couple of providers of hosted open weights inference, due to the complexities of doing so (per Dax from OpenCode's tweets).

> but everyone sees the writing on the wall

i'm... not sure? This assumes ~stagnation in task-possibility. We've had ~exponential progress for like 3+ years now; I'd have never dreamed the tooling I hammer daily would exist in my lifetime just.. 3? years ago. And it's improving daily.

Maybe Open will win, maybe Closed will keep pushing the envelope. The world here is raw enough i don't think anyone can make any significant claim other than 'holy shit this is useful and moving Fast'.

If your doing things the closed models won't let you do; its the whole ball game.

Have you really given GLM 5.2 an honest go ?

So on one hand it states:

Delivery is becoming a sequence of checks, not a ritual The same thing happens when it is time to deliver.

First, I invoke my /definition-of-done skill. It checks whether the implementation covers what was stated in the PR and in the plan. It checks tests and the other validations I care about. It tells me when something looks unusual, broken, or missing. When everything looks good, I invoke another skill, /pr-check-release. That checks the remote PR, updates labels, removes [WIP], adds [RFC], updates the description, and prepares the change for review. If one day passes and nobody on the team reviews the PR, we merge it. The agent also tracks that condition for me, so I can run /pr-merge-dev, and it takes care of the process: merge the PR, delete the remote branch, delete the local branch, and pull dev back into a fresh state. None of these steps is particularly difficult. That is exactly the point.

But then:

I still care about all of those things. Probably more than most people.

Obviously the author has irreversibly became AI-pilled and the day API costs balloon or APIs are down, what work will the author do?

I love using AI but please read the diffs and process them with your human brains and eyes. Spin up your containers manually, test the app, MANUALLY. Talk to real users face to face.

Outsourcing the grunt work is fine, but there's a fine line between that and becoming a button-presser.

> Obviously the author has irreversibly became AI-pilled and the day API costs balloon or APIs are down, what work will the author do?

I am old enough to remember having these thoughts when documentation for things moved from books to being online. I thought looking things up in the internet was a recipe for failure, because the internet was new and unstable and changing, and what happens if you run out of your 10 hours a month of being online, or if someone in your house picks up the phone and you are disconnected?

Clearly those people were internet pilled, and the day the internet costs ballooned or was down, what would they do?

I wrote software before Stack Overflow existed, before Git existed, and before most of the tools I use today existed. I think I will survive an API outage :)

I only wish Opus 4.6 from earlier this year at a faster inference speed. Since Opus 4.6 things have been so much messier and the overall push for more agency isn’t really panning out for agent assisted development as much as they would like

I still use Opus 4.6 (with later models for subagents only sometimes), but I have been preparing for it to go away.

Im talking about the speed of mental model building, understanding concepts, relations and organizational concepts.

Good codebases sort of read themselves. You can guess where things are, how they are sorted and how they work, by understanding and relying on the authors ideas.

“Good” code makes trade offs. While readability is an important constraint, it’s far from being the only constraint. And there are plenty of occasions where objectively better code is subjectively harder to read because other constraints trump human parserability, such as using CPU-friendly memory layouts, SIMD-instructions in tight mathematical loops, and so on and so forth.

Not to mention that readability is entirely dependent on the readers familiarity with particular coding styles. Eg someone unfamiliar with SQL would find ORMs easier to read, whereas I find SQL easier than ORMs. Same is true for any other paradigm, eg for functional vs imperative.

And this is why I hate when people generalise about human readability being the definition of “good code”. For one thing, there will never be a consensus on what is more readable. And external constraints might require subjectively less readable code.

I’ve been plugging away at my running coach style app, powered by the original idea of training for a trail race while living in a flat area with no easy access to natural climbing that has evolved into a fully functional plan generator: https://runcoach.fly.dev

It works well for me so far and I’m pretty happy with it!

Side question: I've always been a recreational runner, running 3/4x a week, completed a few half marathons, and recently decided to _also_ go to the gym to do strength training as it has a lot of benefits for runners too. Should I consider/take creatine, is it useful?

I run and do strength training. I think the consensus is that it's not that beneficial for running. You're going to gain a bit of weight which presumably is going to make you slower.

If you were a competitive runner then you'd probably want to cycle it so that you get the strength training benefit but also optimize for your races.

I take it. I did a 10k race and stopped for two weeks. I'm also not super consistent but I try to take about 5-10g a day.

In terms of optimizing overall health I would say take it + running + strength training is a good combination. The effects are not huge and vary person to person.

Check this video, that gives you a good overview and the description has a good amount of references you can look at: https://www.youtube.com/watch?v=XxZB3Q6EFJU

In short: seems to help with high intensity exercises and post-exercise recovery, helps with muscle development, and a bunch of other benefits.

I’m an ultra runner, I’ll do 50-60 mpw weeks during peak weeks with strength training.

I take 10g creatine, it did wonders for me. More energy and mental sharpness.

Strength training is essential for runners to avoid injury at high mileage. Sleep, strength, and nutrition. It can’t be ignored or you will get injured.

Some folks mention cutting it out to lose weight but at higher mileage I find it hard to keep on weight anyway.

Used to do something similar.

If you intend to have any benefits of going to the gym then protein (and overall calorie consumption) would have to be monitored.

Running sort of conflicts with working out: it wants you to be light and burns plenty of energy.

Gym wants you to gain mass. No mass gains -> gym us useless.

Creatine is kind of an afterthought in this bigger picture. Might give you an extra rep or two but that's it.

Agree on getting enough protein (and sleep, and recovery time).

Strength training isn't all about mass. You can get stronger without gaining mass. There is a neural component and an efficiency component. At some point though you do need a bigger muscle to be stronger. There is also the question of whether you're optimizing for faster (fast twitch) or slower movements.

Strength training is not only about mass but for most intents and purposes it is about gaining as much muscle as possible while not gaining too much extra.

It's not like one can move from squating 100kg to squating 150kg without extra 2-3 kgs of meat.

And there is NO WAY of growing meat in a calorie deficit. And running often leads to that.

Creatine increased my weight and made me a slower runner. I believe for long distance running creatine does not give any benefits. For sprints it might be different. But be aware that if you suddenly increase your training load with the help of creatine, you might easily injure yourself, because your joints are not ready for it.

In a similar situation, I do take creatine. However vo2max will reduce simply due to the waterweight gains.

What if you then use AI to try and maintain only one, a single product into which you’ll put your care and craft to try to make something that’s better than “some dopamine hits”?

That’s how I use it. I might be working on two or three features at a time (iterating, iterating, iterating…), but they’re all scoped and of user value; I don’t feel that I’m just off chasing rabbits.

But I’m also one of those people for whom the “fun” was always solving human problems rather than solving computer problems. I can see how if you are in the latter category AI has already sucked out a lot of joy and how rapidly project switching could be the least-unfun option.

As someone constantly nerd-sniped, the difficulty is that our instincts are still being formed about what this current era of AI tools can and cannot do.

So when a blocker or an idea pops up, it's very easy to use that magic-like tool to solve it quickly and then go back to whatever it's you were doing before.

However, if you care about the quality of your output, that won't be a quick detour. It will pile up with the other "quick" tasks you were doing simultaneously and that's how you end up with 5-10 sessions working on totally unrelated projects.

Shades of https://xkcd.com/1319/

Sure, but for many folks the distraction is irresistible. It was difficult already to put care and craft into a product, having a slot machine for your attention makes it damn impossible.

That's funny, that's the exact conclusion I'm starting to come to

Having tried many of these hosting services to host/play with toy apps, DigitalOcean and Fly.io are both unparalleled GOATs.