I can only talk about GLM 5.1 which is roughly at sonnet 4 levels imo.

It's good, does most tasks well that I throw at it, but will fail at anything congitive/complex. It gets stuck often. It costs ~6$ a month though

The Kimi problem is it doesn’t follow instructions and goes off track often.

Other than that it’s pretty decent (for the price).

price/token isnt the only thing relevant. if you have to ask the AI again, it'll cost you more than when it gets things right in the first place.

so better models may still be cheaper even if the price per token is higher.

> My theory is that US enterprise just can't send data to Chinese

Lots of US providers are hosting these “open source” models so doubt that’s the problem.

I want Opus to be only marginally better, but I do mostly research engineering and its ability to not fuck up my projects is absent. Every time my credits lapse I let kimi and composer2.5 have some play and it’s basically just an excuse for me to keep playing computer because when the oai/ant credits refresh I always need to spend hours recovering from the other models either misconceptions or boneheaded eng practices. Even when I only let it touch my web games…

Is Deepseek just eating cost or are people able to host their open models for comparable costs?

These models have open weights, but at the moment most flagship models are practically accessible only through third-party model providers. The main exception is models in the ~30B parameter range, which can still be run on consumer-grade GPUs. That said, even consumer GPUs have become increasingly expensive and difficult to justify in recent years.

I keep trying to switch to the Chinese models, but I keep finding myself asking Claude to fix their outputs. (Both functionality and style.) So I always end up switching back.[0]

I also keep trying GPT, which is quite solid. Very fast, great at debugging. But its code is often overly clever and hurts my brain.

(Maybe fixable with prompting. I tried and it helped the Chinese ones a bit. Just tell them do be elegant, like in the old image AI days "+good -bad"!)

For now I do still need my human brain to actually be able to make sense of the stuff, and Claude is the only one that consistently meets that requirement.

But I am hoping that one of these days, one of the Chinese labs figures out the special sauce :)

--

[0] (For smallish edits, though, I am having a great time with DeepSeek Flash. Practically unlimited AI on tap! How cool is that.)

I use glm5.1 plus pi with a few customized skills and am very happy with it. I hadn’t touched my Claude 5x plan for a couple of weeks but opened it back up in Claude code when fable was released and did a few tasks and still was happy to return to glm/pi.

It seems tacked on pretty quickly - I would have expected they try a little more legalese regarding what counts as a "user interface".

I find that I don't use a ton of output tokens. I'm usually around 95% cached input, 4% input, and 1% output.

For me, the big thing with MiMo-V2.5-Pro and DeepSeek V4-Pro is that cached inputs are practically free. Kimi K2.7 Code is 53x more expensive for cached inputs which is 95% of my costs.

If I use 95M cached input tokens, 4M input tokens, and 1M output tokens, that'd be: $18 for cached input on Kimi K2.7 Code vs $0.34 with MiMo/DS; $3.80 for inputs on Kimi vs $1.74 with MiMo/DS; and $4 for output on Kimi vs $0.87 with MiMo/DS.

Of all the places where I'm accumulating costs by using Kimi, it's the cached inputs. The real savings with MiMo/DS's price cut is the cached inputs.

This is the cursor callout.

Don't make us shame you into disclosure

> they asked you to 'advertise' them basically.

To be clear, the “advertising” clause just requires you to disclose that you use the thing somewhere in the product, such as credits in an “About” section.

I've kind of given up on the routers for "free" inference, as you would expect, they tend to give you sub-par thinking because they are obviously trying to conserve as much inference as possible.

I've had some success turning my macbook M1 pro into a heating pad with Qwen 3.6 35B A3B MTP. Trying to use Gemini models "locally" resulted in a similar "short shrift" of effort resulting in mistakes and lots of turns. The reports of Fable being relentlessly "proactive" shows you can go the other direction as well, if you have strong enough branding and effective invoicing.

Deepseek V4 is far from Opus 4.6 level, it might look like it at first glance, but the general reasoning (especially multi-steps) is frankly far off. It's good enough to build great things don't get me wrong, but there is really something that is different from Anthropic models.

What do you mean by custom format? Non-json?

Would be nice to have 5.2 and 4.6 for comparison.

Check out TNG on huggingface

They are a consultancy in Germany, but I watched a presentation on them tuning and removing bias from Deepseek models. It was quite interesting.

https://www.tngtech.com/en/about-us/news/release-of-deepseek...

(I upvoted your question as I agree)

Its not just code we need to worry about, its also subliminal messaging and other things.

> Has anyone taken these open weight models from China and stripped the CCP out of them?

The CCP is not influencing my Rust code quality that much. Though I did notice all my lifetimes are now 'static because nothing is ever allowed to leave the party's ownership, unsafe blocks require approval from a central committee.

Honestly the scariest part is that shared mutable state is forbidden unless the state is doing the sharing.

Otherwise it is pretty ok.

In my experience, there's little difference between implementing individual functions between frontier models and SotA ~30B param models.

Once you have a coherent design (the hard part), you can feed it to a pretty small model and get basically the same quality.

They'll not one-shot, but they're faster and cheaper, so it still works out in your favor.

Plus you can do it locally...

I really hope we stop using the term "Chinese models". It has this air of Negative connotation. It's the equivalent of calling cars Japanese, which people used to do but now is almost entirely meaningless. You just call them Toyota, Honda, Lexus etc.

The difference in outcome isn't that big but yes, you need to be more rigorous. For instance I've found that the Kimi K2.5 and K2.6 models will comment out failing tests rather than fix a problem they just caused (mistaking them for "pre-existing failures"), so you need to specifically make commented-out tests break the build. I've not personally had that problem with any of the Anthropic or OpenAI models.

The model is natively quantized (i.e. it was trained that way in the first place, so this is not a post-training quantization which degrades performance).

Eh even corporate created LLMs are suspect to corporate biases. Nothing is safe.

Sounds like something that heretic or similar might be useful for?

https://github.com/p-e-w/heretic

It's not 2.7. It's 2.7-Code, and it's 2.6 token-optimised for coding.

https://platform.kimi.ai/docs/guide/kimi-k2-7-code-quickstar...

I all it advertising clause, because I remember still in the 2000s seeing an Apple ad which at the end of it showed "Unix" or something like that on it, and I remembered that was one of the BSD license requirements, or maybe Apple just did it also just to proudly boast using Unix.

> The reports of Fable being relentlessly "proactive"

For the curious: https://news.ycombinator.com/item?id=48498573 - “Claude Fable is relentlessly proactive”.

> I've kind of given up on the routers for "free" inference, as you would expect, they tend to give you sub-par thinking because they are obviously trying to conserve as much inference as possible.

Xiaomi MiMo ($6/mo: https://platform.xiaomimimo.com/token-plan) & Alibaba Qwen ($50/mo: https://www.alibabacloud.com/en/campaign/ai-scene-coding) have generous limits on fixed subscriptions.

Tangent: did the MTP help you at all? I’ve tested that model back to back on my M1 Max MBP and the MTP version was actually marginally worse. I wonder if I didn’t use the right settings, although I tried several based on the obvious sources.

Ah is that what it is? I don't use Cursor, never saw it as being relevant to me, but would not surprise me.

Wasn't the end of that story that Cursor had a non-disclosure licence, so they had not done anything wrong towards Moonshot?

Cursor had a specific licensing agreement that allowed them to brand it how they want.

Shaming others when all AI is trained off scraped content and code huh? Many of those sources either breaking ToS or being illegal, such as Anna’s Archive. Bold move. And Chinese models in particular have been accused of distilling off American models.

Don’t you know there’s no honor among thieves?

For me, it has a positive connotation! In my experience, Chinese Model means cheaper, but still quite effective model you can use for millions of tokens without burning your entire wallet in seconds. That's why I get more excited over a Chinese model release over American models.

I have a similar experience. However, when including code review, I think the GPT model is the most impressive

Japanese cars is actually a positive qualifier. I'd say anything Japanese motor-powered.

No thanks.

The term seems to have the connotation of "competitive at 1/10 the price of Claude", so I don't see the problem.

It's not Harbor Freight Chinese (and heck even they have decent stuff sometimes now too).

You don't think people still talk about Japanese cars as a distinction in quality from US or European ones?

I don't know, I tried using one of the Chinese models and it was VERY quick to scan my entire home dir, so maybe your threat surface is a little different than mine

I don't think "Chinese" is pejorative in this context any more than "American" is. They are one of the two ecosystems. What's wrong with saying "Japanese cars" today?

They are all funded and owned by the same entity, the CCP, so it probably would be better to call them CCP models.

Edit: Downvoting something doesn't make it false.

You are right. I agree.It may seem like a kind of bias, but I hadn't thought of that part. Thank you for pointing out my bias.

I wonder why it's the natural tendency of models to BS or do stuff like this when they don't have the correct answer - it's clear that they can program refusal into them, but for some reason, refusal has to be injected after the fact, and models can't really arrive at the conclusion that they can't answer properly.

Isn't it not completely quantized? I thought there were some dense parts but most is int4?

Everything is the same is not a serious argument because they are not the same.

But the huggingface link mentions BF16, F16, and I32?

Moonshot licenced it to Fireworks AI who licenced it to Cursor.

> Cursor had a specific licensing agreement...

Cursor had an "agreement" with Fireworks.ai, which apparently allowed them to RL Composer 2 atop Kimi Base 2.5 without attribution: https://x.com/Kimi_Moonshot/status/2035074972943831491 / https://archive.vn/CcdkI

Composer 2 performed differently on evals than Moonshot.ai's coding models: Cursor claims theirs is better than Claude Opus 4.6: https://x.com/fynnso/status/2034706304875602030 / https://archive.vn/bVtik. And, per Lee Robinson (Cursor employee), it is very likely Cursor builds its own foundational model for Composer 3.

Cursor's composer models are finetuned kimi

Models can't scan anything.

They return instructions for you to do something, and you or a script you permit chooses to execute what the model tells you and return the result to the model.

Do you believe there's some meaningful benefit to the American VC funding model in this case? It's not clear to me what you're trying to say or why you think it's an important distinction.

For those that don't like calling them CCP models, may I remind you, the CCP won't let Chinese AI researchers out of the country any more without securing approval first[1].

[1] https://www.tomshardware.com/tech-industry/artificial-intell...

Sadly there is a pejorative context. The constant us, the free world vs China, the evil Soviets rhetoric from every major news establishment and executive creates that negative view

> What's wrong with saying "Japanese cars" today?

Only that it’s a fairly meaningless grouping. When japan first entered the car market in north america there might have been some commonality, but now what characteristics do they share that some american cars don’t have? They’re not even imported a lot of the time.

Given that, it does start to feel tinged with racism if someone insists on grouping things together that don’t really belong together.

As for Chinese LLMs, the term doesn’t “feel” pejorative to me - but i also don’t see a totally clear set of attributes they share. Not all are open-weight. Some are small and can be run on consumer hardware, some are huge. They even have a variety of answers to what happened june 3rd 1989

I tend to agree with the comment in my reply thread about whether we really need to add biased modifiers to the essence of a good product. I think every national system in this world is flawed. And in this context, 'China or Chinese' is often used in a negative sense, like 'Made in China'. But KIMI is a good model, and I think the comment that pointed this out to me correctly identified my unconscious bias.

And even if the Chinese Communist Party provided funding, the result is still transparently released. So even if it is some kind of propaganda, I don't see what the problem is.

Is the monopolistic greed of American companies 'good', and China's greed 'bad'? I do have that question.

I've heard this claim before but I've never seen any evidence.

I assume it's a lack of care when RLing them.

RL has a tendency to reinforce cheating when the cheats are easier to find than the final solution.

So when making your RL environment, you need to spend a lot of effort on finding ways the model can cheat and penalizing them.

Not every weight is quantized. For example, those weights which don't take much space or are highly important are left in higher precision. State-of-art quantization of weights is never done uniformly (i.e. to all weights and in the same way).

I don't believe safetensors has a native int4 dtype, so they packed 4 int4s into a bf16 in this checkpoint.

"You're absolutely right"?

yes, yes, the spectre of communism, BYD is the CCP, Alibaba is the CCP, stealing your children and eating them for Mao, bla bla bla.

I have a feeling you'd be slightly salty at people saying "Google and Tesla are making CIA models"

They are unusable (unless you want to deliberately destroy your codebase). So if Cursor's models are Kimi based, then well. I'll skip them altogether.

On the other hand the Trump administration has successfully managed to make Chinese seem better than American, so there might not be that much of a pejorative context any more..

> but now what characteristics do they share that some american cars don’t have?

Better overall design?

The question is not whether it is a good model, it is whether the model can be trusted to not act intentionally maliciously against certain topics or certain users.

We live in a time of a great geopolitical rivalry and high tensions with an emergent technology with tons of national security implications. To pretend otherwise is silly, and to fail to ask the question, dangerous.

> now what characteristics do they share that some american cars don’t have?

Typically the answer is "reliability", which is a positive trait, which makes the original callout about negative connotations very odd to me.

"You hit the nail on the head" LOL

Whether or not it's propaganda is different from the fact that it is owned by the CCP.

Have you looked, or you’re just waiting for someone to hand it to you?

Assuming you are just naive like so many others about China...

China is a communist country with elements of capitalistic markets baked in. But the capitalistic elements are mostly a facade. Underneath, the state retains full ownership and control of all business. The CCP runs all aspects of the government (including the courts/judges), and is the single entity that decides what directions the country (and it's businesses) will move in.

The CCP, who defacto owns everything and has ultimate final say on everything, has one leader that has the ultimate final say on _everything_, Xi Jinping.

So while the waters of CCP models feel warm and free, understand it's not organically like that.

I'm using Composer extensively, and it works great for me. Your experiences are not universal.

Kimi works great in their CLI, but their CLI has a number of workarounds for quirks of their models, including detecting when the model gets into a loop, and reverting to a checkpoint but letting the model compose a "message" to its past self (search their CLI for "BackToTheFuture"...) It doesn't work so well in a harness that doesn't take those quirks into account.

They are far from unusable. They aork great for 80-90% of a typical full stack dev. Alot less useful for more noche stuff

I wouldn't skip at least testing the original. Model distilling done by Cursor could be the culprit.

Composer 1.x was poor. The new one is a totally different beast and absolutely fine for day to day.

They're not unusable, they're just bad when compared with all the real frontier models.

I only use composer 2.5 day to day and it works fine with human review.

I mean...

Since its development, IQT has invested in over 750 startups spanning diverse technological sectors, including:

  - Artificial Intelligence
  - Space Technologies
  - Microelectronics and Quantum Computing
  - Life Sciences
  - Cybersecurity
  - Hardware
  - Energy

This broad portfolio has enabled IQT to address a wide array of national security challenges while supporting the growth of innovative startups…

https://en.wikipedia.org/wiki/In-Q-Tel

https://www.npr.org/sections/alltechconsidered/2012/07/16/15...

https://www.cgai.ca/th_bn_iqt

Going by their response you appear to have been correct lol

Google and Tesla making products to sell to the government is different than the government funding the government to make products for the government.

In China it's all one entity with these mock facades of privatization. Trump cannot instruct Google to put picture of dogs on their homepage. If Xi wakes up and wants dogs on Alibaba's homepage, give it 30 minutes.

It's wholly ignorant or dishonest to make the comparison.

You're right, but the bias in the US certainly persists. "China = bad" is an assumption that many people still make without any self-reflection about the ways in which the US is now at least as bad.

> The question is not whether it is a good model, it is whether the model can be trusted to not act intentionally maliciously against certain topics or certain users.

We absolutely know that we can't trust the American model not to do that - it's "by the oligarchs, for the oligarchs" - so it's not clear what the claim really is.

Chinese AI models also share a positive trait: they offer more bang for the buck.

Doesn't matter, because they're open-weight, so I can just download them to my PC and... hey, look, now they're owned by me! Unlike the "good" Western counterparts which are all fully proprietary. (Except Mistral, but they're nowhere near SOTA.)

> China is a communist country with elements of capitalistic markets baked in.

While I get the point you're making (it should be pretty obvious to anyone who's held a newspaper), I think it's important regardless to point out that Chinese companies AFAIK aren't worker-owned or -controlled, so you can't exactly call it communism, either. And they obviously do not have a "free market capitalism", as you just discussed.

It's simply a highly authoritarian state then, I guess?

The burden of evidence is on the accuser.

Crazy mental gymnastics if you think the American oligarchs don’t have the final say on everything in America. They’re just smart enough to do it behind the scenes, well they used to be. They barely bother anymore.

I'm not salty, they are just confused about the difference between free enterprise capitalism and communism, which is understandable.

I know, but it's far too fun to bait the parent into revealing how ignorant they are.

> Trump cannot instruct Google to

Tim Apple and the other tech CEO constantly groveling at Trump’s feet indicates that he might be able to do that.

Just like threatening TV networks about having their licenses revoked of blocking mergers unless they fire the people making fun of him on TV (of course with slightly mixed success)

> Trump cannot instruct Google to put picture of dogs on their homepage.

Sundar Pichai would personally be barking on a livestream on the homepage.

Trump is quite literally the one president showing that the US has zero rules or anything to hold power back from the white house, really not the example you want.

What is hidden in the weights matters.

Its akin to the accusation that the UK is a parliamentary system...

The companies are all worker owned, because the state exists for the people, and the state owns everything. At least on paper that's how it is sold. After all it is the Peoples Republic of China.

Generally I consider conspiracy to be the "crazy mental gymnastics"

Seems like everyday Trump has another order struck down by the courts.

Sundar can do whatever he wants, but he has no legal obligation to do any of it.

Ah yes, those pesky Chinese backdoors that no single instance was ever found, even though Chinese open-weight model are a thing for many years now. Many people burn through millions of tokens on these models every day - surely someone would have triggered one of those backdoors, right?

Or that pesky CCP censorship and propaganda baked into the model, which any random guy can remove from whichever model they want as a single weekend side project with an off-the-shelf tool[1]. (Try it. It's fun. I've done it myself.)

[1]: https://github.com/p-e-w/heretic

I mean, if that's the bar then the state owns everything in America as well. After all, you don't really own your land if the state can regulate what you do on it, what it can be used for, what you can build on it, and can take it away if they really need it. The state owns the land, the money supply, and regularly restricts and instructs businesses to take or desist from actions.

As such, the state owns everything in both countries, the only differences are to what extent they control things.

I wouldn't even call the USA a capitalist system anymore, the economy is so heavily regulated and interfered with. It's a "managed economy", like pretty much every other nation's economy in the present day.

Courts and legal obligations are to a certain extent irrelevant at this point. There are plenty of illegal ways that Trump can fuck over Google and face no consequences.

e.g. he had Colbert fired (and who knows what else) by threatening to block the Paramount/Skydance merger

> Trump cannot instruct Google to put picture of dogs on their homepage.

I'm sorry, but that was a horrible example. Corporations have no obligation to donate money to the ballroom yet Google has donated millions.

I agree it is an empirical question. I do not know if that research has been done in the open sphere. But please, do not pretend that there isn't a real geopolitical rivalry going on that makes such questions a legitimate, non-fruity concern.

In the US you can take the state to court and win...

>Corporations have no obligation to donate money to the ballroom yet Google has donated millions.

Imagine living in a country where they have the obligation.

This is a fair point, alongside the one about the hidden content in the weights.

Exactly why my prime suspect would be the one country with focus on proprietary models, and the one country prone to bombing others, including with nuclear weapons.

Sure, but the difference is that one side (Anthropic, OpenAI, Google and co.) hoards everything, keeping it proprietary behind API paywalls and constantly spewing AI doomer rhetoric while limiting what you can do "for your own safety" (especially Anthropic; Dario has been consistently doing this since GPT-2 days, every time claiming that things are "too dangerous" for the common folk to handle). While the other side (big, bad China) releases all SOTA open-weight models with which you can do whatever you want with, along with a ton of open research.

So yes, there is geopolitical rivalry, but one side is deliberately antagonistic (not releasing anything in the open, putting arbitrary restrictions, spewing toxic rhetoric, applying sanctions, etc.) while the other side is letting everyone (including their rivals) to use what they've produced with little-no-to restrictions.

I'm under no illusion that if the situation was reversed China would most likely do the same, but as things stand you can probably guess which side I'm rooting for here (at least until the roles reverse).

Yes, each are following their own business strategy, frontier labs have no incentives for releasing open weights, while second and third-tier labs, it is one of their few plays to gain market/mind share. But business is only part of it, as national security is another. It may be that the CCP has been relatively hands off exactly because of my concern, judging that market share and reputation is more important (for now).