For now. That doesn't really change the risk, that just means they are all hyper competitive right this moment, and so they are comparable. If one of them becomes king of the hill, nothing stops them from silently degrading or jacking prices.
The only shield is to not be dependent in the first place. That means keeping your skills sharp and being willing to pass on your knowledge to juniors, so they aren't dependent on these things.
Of course, many people are building their business on huge AI scaffolding. There's nothing they can do.
It's similar in the way many businesses transitioned their scalability etc. to 'the cloud' starting a couple of decades ago.
It's a combination of loss of control and abdication of responsibility. They can claim to the customer the reason the service went down is now Microsofts or Amazons etc. etc. fault. Ultimately the end-user was the one that ended up losing.
It was a choice. There was something they could do - and keep everything in house, although cost-competiveness becomes an issue at some point and you get priced out of your target market. Everyone loses except for the cloud computing (or now AI) providers.
But, so far, competition remains fierce. Anthropic still has the best tools for writing code. That lead is smaller than it's ever been, though. But, honestly, Opus 4.5 is when it got Good Enough. If Anthropic suddenly increased prices beyond what I'm willing to pay, any model that gives me Opus 4.5 or better performance is good enough for the vast majority of the work I do with agents. And, there are a bunch of models at that level, now maybe including some discount Chinese models. Certainly Gemini Pro 3.1 is on par with Opus 4.5. Current Codex is better than Opus 4.5 and close to Opus 4.7 (though I won't use OpenAI because I don't trust them to be the dominant player in AI).
I often switch agents/models on the same project because I like tinkering with self-hosted and I like to keep an eye on the most efficient way to work...which models wastes less of my time on silly stuff. Switching is literally nothing; I run `gemini` or `copilot` or `hermes` instead of `claude`. There's simply no deep dependency on a specific model or agent. They're all trying to find ways to make unique features for people to build a dependence on, of course, but the top models are all so fucking smart you can just tell them to do whatever thing it is that you need done. That feature could probably be a skill, whatever it is, and the model can probably write the skill. Or, even better, it could be actual software, also written by the model, rather than a set of instructions for the model to interpret based on the current random seed.
Currently, the only consistent moat is making the best model. Anthropic makes the best model and tools for coding, but that's a pretty shallow moat...I could live with several other models for coding. I'll gladly pay a premium for the best model and tools for coding, but I also won't be devastated if I suddenly don't have Claude Code tomorrow. Even open models I can host myself are getting very close to Good Enough.
They won't ever be SOTA due to money, but "last year's SOTA" when it costs 1/4 or less, may be good enough. More quantity, more flexibility, at lower edge quality. It can make sense. A 7% dumber agent TEAM Vs. a single objectively superior super-agent.
That's the most exciting thing going on in that space. New workflows opening up not due to intelligence improvements but cost improvements for "good enough" intelligence.
Why should anyone waste time on poorer results? I'd rather pay my $200/mo because my time matters. I'm not a poor college student anymore, and I need more return on my time.
I'm not shitting on open weights here - I want open source to win. I just don't see how that's possible.
It's like Photoshop vs. Gimp. Not only is the Gimp UX awful, but it didn't even offer (maybe still doesn't?) full bit depth support. For a hacker with free time, that's fine. But if my primary job function is to transform graphics in exchange for money, I'm paying for the better tool. Gimp is entirely a no-go in a professional setting.
Or it's like Google Docs / Microsoft Office vs. LibreOffice. LibreOffice is still pretty trash compared to the big tools. It's not just that Google and Microsoft have more money, but their products are involved in larger scale feedback loops that refine the product much more quickly.
But with weights it's even worse than bad UX. These open weights models just aren't as smart. They're not getting RLHF'd on real world data. The developers of these open weights models can game benchmarks, but the actual intelligence for real world problems is lacking. And that's unfortunately the part that actually matters.
Again, to be clear: I hate this. I want open. I just don't see how it will ever be able to catch up to full-featured products.
The trick is going to be recognizing tasks which have some ceiling on what they need and which will therefore eventually be doable by open models, and those which can always be done better if you add a bit more intelligence.
When was the last time you used any of them? Because, a lot of people are actively using them for 9-5 work today, I count myself in that group. That opinion feels outdated, like it was formed a year ago+ and held onto. Or based on highly quantized versions and or small non-Thinking models.
Do you really think Qwen3.6 for a specific example is "50%" as good as Opus4.7? Opus4.7 is clearly and objectively better, no debate on that, but the gap isn't anywhere near that wide. I'd call "20%" hyperbole, the true difference is difficult to exactly measure but sub-10% for their top-tier Thinking models is likely.
So the starting point is Opus 4.7 pricing and we're contrasting alternatives near the top end (offered across multiple providers).
Also I said 20% was hyperbole, meaning far too high.
Those closed weight models aren't available like we're discussing. They're only available from the vendor that created them.
Sure, we use Google Drive, too, but that's just for sharing documents across offices, not for everyday use. For that, the open source model is a clear winner in my book.
The breakeven at this price is 6 minutes of productivity per work day for an engineer making $200k.
Are you suggesting that someone making $20k should be spending $200/mo on Claude?
If you pay someone $20,000 for labor, and they save 65 minutes worth of labor per day using a $200/mo Claude subscription, you are better off buying the Claude subscription.
But the specific numbers in my prior comments aren’t really relevant to my point. Adjust for whatever numbers you want.
I've done some napkin math and CC code makes me more efficient when I pay 200/ month, but it wouldn't if I had to pay api prices
The math is highly in favor of us using it at our company and we are paying API pricing. I don’t imagine there’s a lot of people using Claude without getting their money’s worth…?
I would start claude in Yolo mode, tell it keep trying new ideas until it runs out of 1m context. (Every day I am giving it a hint to explore different directions as the sessions before)
Twice a day for a month, fits well into CC max plan.
I guess if I had to pay per token I would still use it but only for tasks where the value is clearer and immediate.
You've got the real insight with this claim.
This is the way the world is moving. Open source isn't even going where the ball is being tossed. There is no leadership here.
You're spot on.
If the cost to deliver a unit of business automation is:
A. $1M with human labor
B. $700k human labor + open source models
C. $500k human labor + $10,000 in claude code max (duration of project)
D. $250k with humans + $200k claude code "mythos ultra"
Your poor college students and hobbyists will be on option "B". But this won't be as productive as evidenced by the human labor input costs.
Option "C" will begin to disappear as models/compute get more expensive and capable.
Option "A" will be nonviable. Humans just won't be able to keep up.
Open source strictly depends on models decreasing their capability gap. But I'm not seeing it.
Targeting home hardware is the biggest smell. It's showing that this is non-serious, hobby tinkery and has no real role in business.
For open source to work and not to turn into a toy, the models need to target data center deployment.
The real money in this market, though, is going to be made in the C suite, and they don't really care about the model. They don't care if it's open source, closed source, or what it is. They don't want to buy a model. They're interested in buying a solution to their problems. They're not going to be afraid of a software price tag -- any number they spend on labor is far more.
Labor is something like 50%+ of the Fortune 500's operating expenses -- capturing any chunk of this is a ridiculous sum of money.
Wanna bet on it?
Who said so? GLM 5.1 is 90% Opus, at least. Some people quite happy with Kimi 2.6 too. I did not try Deepseek 4 yet but also hearing it is as good as Opus. You might be confusing open source models with local models. It is not easy to run a 1.6T model locally, but they are not 50% of SOTA models.
Because in almost no real-world project is "programming time" the limiting factor?
This kind of rhetoric is not helpful. If you want to make a point, then make one, but this adds nothing to the conversation. Maybe open source models don't work for you. They work very well for me.
I'm not disagreeing per-se but if you think the benchmarks are flawed and "my real world usage" is more reflective of model capabilities, why not write some benchmarks of your own?
You stand to make a lot of money and gain a lot of clout in the industry if you've figured out a better way to measure model capability, maybe the frontier labs would hire you.
The gap has been shrinking with each release, and the SOTA has already run into diminishing returns for each extra unit of data+computation it uses.
Do you really want to bet that the gap will not eventually be a hairs breadth?
Edit: the replies to my comment are great examples of what I’m talking about when I say it’s hard to determine what hardware I’d need :).
Starting closer to 40k if you want something that's practical. 10k can't run anything worthwhile for SDLC at useful speeds.
(If you are willing to let the machine work mostly overnight/unattended, with only incidental and sporadic human intervention, you could even decrease that memory requirement a bit.)
Also, I don't know of a general solution to streaming models from disk. Is there an inference engine that has this built-in in a way that is generally applicable for any model? I know (I mean, I've seen people say it, I haven't tried it) you can use swap memory with CPU offloading in llama.cpp, and I can imagine that would probably work...but definitely slowly. I don't know if it automatically handles putting the most important routing layers on the GPU before offloading other stuff to system RAM/swap, though. I know system RAM would, over time, come to hold the hottest selection of layers most of the time as that's how swap works. Some people seem to be manually splitting up the layers and distributing them across GPU and system RAM.
Have you actually done this? On what hardware? With what inference engine?
Hooking up Claude Code to it is trivial with omlx.
So you can run 1 agent locally on $1k to $3k hardware
They can run a fleet of thousands
Yes, it's possible to run tiny quantized models, but you're working with extremely small context windows and tons of hallucinations. It's fun to play with them, but they're not at all practical.
Practical? Maybe not (unless you highly value privacy) because you can get better models and better performance with cheap API access or even cheaper subscriptions. As you said, this may indefinitely be the case.
Yes, a lot better, but still terribly unreliable and far less capable than the big unquantized models.
[†] The latest Qwen 3.6 whatever has been a noticeable improvement, and I'm not even at the point where I tweak settings like sampling, temperature, etc. No idea what that stuff does, I just use the staff picks in LM Studio and customize the system prompts.
Until very recently, local models been little more than brittle toys in my experience, if you're trying to use them for coding.
But lately I've been running Pi (minimal coding agent harness) with Gemma4 and Qwen3.6 and I've been blown away by how capable and fast they are compared to other models of their size. (I'm using the biggest that can fit into 24gb, not the smaller ones.) In fact, I don't really need to reach for Claude and friends much of the time (for my use cases at least).
Competition (OpenAI vs Anthropic is fun to watch) and open source will get us there soon I think.
AI tools... do what you already do, sometimes faster, sometimes worse, usually both depending on the task.
There's a massive gap of necessity between them.
Not the best argument.
Also there is nothing without dependencies. Loose coupling means coupling.