Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling • Hacker News Board

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#111

Post by m_ke » Mon, Jul 20, 2026, 5:03 PM UTC

cmrdporcupine wrote: post #23

energy123 wrote: post #13

Earlier quoted context omitted.

None of my use cases require frontier capabilities but I still pay $200/month to a frontier lab. I value the additional time saved at more than $200/month. If I had to pay actual API rates, then I'm not sure what I would do, but it would not be an easy decision.

Yes, the reckoning here will happen in a year or two when the (probably subsidized, maybe?) coding plans become either unavailable or much more costly. It's also already the case that in larger companies people do not have access to these plans as employees and must use API rates. There's also the political angle. When Anthropic and OpenAI held back their top of the line models because of Bessent & Trump's bullshit,…

k3 costs will go down at least 3x within a week of the weights dropping.

we'll get new quants, dspark speculators, distills and optimized kernels

as long as there are near frontier models available there will be inference providers selling them at or below cost of inference in attempt to get market share.

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#112

Post by surgical_fire » Mon, Jul 20, 2026, 5:05 PM UTC

simianwords wrote: post #22

To everyone praising Open weight models, could you answer a simple question? If Anthropic doesn't make money because of distillation attacks, how would they convince investors to invest in them, such that it makes financial sense for Anthropic to train even bigger models? Assuming it is preferable for everyone that we get better models in the future. Distillation attacks remove the financial incentive.

[Pending evidence that Chinese models are only based on distillation]

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#113

Post by bko » Mon, Jul 20, 2026, 5:06 PM UTC

jhonof wrote: post #28

bko wrote: post #5

I think the risk is overstated. For one, on the margin people are willing to pay a lot for slightly better models. I know personally the value the LLM adds to my workflow is considerably more than the $200/m I pay the frontier labs. I have no interest in optimizing that to get it slightly lower. There are a very vocal minority that optimizes this or companies whose LLM expense is marginal, but I think that's the mino…

> For one, on the margin people are willing to pay a lot for slightly better models. I know personally the value the LLM adds to my workflow is considerably more than the $200/m I pay the frontier labs. I have no interest in optimizing that to get it slightly lower. There are a very vocal minority that optimizes this or companies whose LLM expense is marginal, but I think that's the minority (correct me if I'm wrong,…

> Claude and OAI are not valued at $1T because of their harnesses

They're valued at that because they add a lot of value and people pay for the product. The product is more than the LLM. If you want argue the value of the harness vs LLM but flippant remark adds nothing.

> Enterprise excel is like $50 a month, the closed source labs charge orders of magnitude more than that per user per month for enterprise, and want to charge more if they can to maintain their valuation. I spent $20 in an afternoon yesterday using my work plan for one workflow for reference, I probably spend at least $50 a day and I don't even do that much coding anymore.

I want to know what percent of their revenue is this API spend compared to just a Claude/Codex max plan that's $200/m and very generous.

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#114

Post by overgard » Mon, Jul 20, 2026, 5:07 PM UTC

onlyrealcuzzo wrote: post #49

overgard wrote: post #14

I keep thinking about the Figma thing. If you're unaware, here's the google summary: ---- The Board Departure: Mike Krieger, Anthropic’s CPO and a co-founder of Instagram, sat on Figma’s board of directors. He resigned on April 14, just days before news of Claude Design broke. This sparked speculation over conflict of interest and the use of proprietary product strategy information. Betrayal of Partnership: The launc…

> I would suggest to people using LLMs: you should be cautious about giving these companies data or relying on them. If you're building an AI startup, there's a very good chance they could decide to directly compete with you if your idea has traction. You're also at their mercy for API pricing etc. LLM generated code is not copyrightable, so even if they do "steal" it - I don't think there's legal grounds to do anyth…

So, code is not the only data they potentially have access to, and it might not even be the most interesting data. There's also whatever other things you plug the AI into: slack, email, internal documentation, etc.

Also, a lot of people using LLMs are using it on code bases that were written before LLMs were common for coding. So in that context, you really are allowing them to train on your code for free.

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#115

Post by bko » Mon, Jul 20, 2026, 5:08 PM UTC

esperent wrote: post #12

bko wrote: post #5

I think the risk is overstated. For one, on the margin people are willing to pay a lot for slightly better models. I know personally the value the LLM adds to my workflow is considerably more than the $200/m I pay the frontier labs. I have no interest in optimizing that to get it slightly lower. There are a very vocal minority that optimizes this or companies whose LLM expense is marginal, but I think that's the mino…

> considerably more than the $200/m I pay the frontier labs. This is a very rich / developed country privilege perspective. Where I live, it's not unusual for a monthly wage to be around $200. Of course developer wages are much higher, maybe as much as $1000 a month, but $200 is still a huge chunk of that so it doesn't really matter how much "value" you get out of it if you're no longer able to pay rent or buy decent…

Thats great, but the market will always be driven by rich developed countries. I don't think Anthropic and OpenAI is banking on third world adoption in their growth estimates.

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#116

Post by quotemstr » Mon, Jul 20, 2026, 5:08 PM UTC

LarsDu88 wrote: post #48

The open weight, open architecture releases of the past several days has me more convinced that ultimately, the winner will be whoever burns their models to ASICs fastest. The LLMs themselves are capable of doing some aspects of chip design as evinced by the K3 press release. Furthermore, the frontier models are "good enough" for a wide swathe of tasks and will soon hit that threshold for a good amount of software en…

What precisely is an ASIC supposed to do that a more programmable accelerator can't? Memory latency is memory latency: doesn't matter whether it's embedded in a cache-line wait state or some flipflop state machine. ROM isn't going to be faster than RAM either.

Likewise, for compute, is the ASIC somehow going to beat a systolic array? You can't have one circuit per weight: the die area and electrical fan-out would be insane. I'm not seeing how an ASIC specialized for a specific model would actually help much.

I mean, sure, we can build more specific accelerators, e.g. for softmax, but these work fine in the context of a programmable pipeline.

Yes, there are more exotic things out there, like optical matrix multiplication systems. Those are different. But above, aren't you talking about just doing conventional digital linear algebra, but with a model-specific set of circuits?

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#117

Post by bko » Mon, Jul 20, 2026, 5:08 PM UTC

kgwgk wrote: post #11

bko wrote: post #5

I think the risk is overstated. For one, on the margin people are willing to pay a lot for slightly better models. I know personally the value the LLM adds to my workflow is considerably more than the $200/m I pay the frontier labs. I have no interest in optimizing that to get it slightly lower. There are a very vocal minority that optimizes this or companies whose LLM expense is marginal, but I think that's the mino…

> a huge value is the […] Codex harness. There are open source implementations Like Codex https://github.com/openai/codex

If you think Codex is some neutral harness designed to work equally as well for all LLMs, I got a bridge to sell you...

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#118

Post by quotemstr » Mon, Jul 20, 2026, 5:09 PM UTC

akiselev wrote: post #101

stillpointlab wrote: post #73

Earlier quoted context omitted.

I'm not convinced, mostly because things like crypto, which I believe went into ASICs, were based on very slowly moving and mostly understood algorithms. LLMs and model architectures seems significantly more volatile. I wouldn't want to be working out the finer details of my chip rollout only to find a new paper/approach that give multiples of performance. So I guess it depends on how much the latest-greatest model m…

The tokens per second performance numbers coming from Cerebrus/Talas are several orders of magnitude higher than models running on GPUs, which is such a huge step change that it will enable many more uses of LLMs that are impractical otherwise. I.e. think about gamers and burning in an LLM chip on a game console like a future Play Station - it doesn't matter if its a frontier LLM if it allows them to talk to in game…

> Cerebrus/Talas

They are fast, but they're still programmable accelerators, not a model burned into the gates.

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#119

Post by athrowaway3z » Mon, Jul 20, 2026, 5:11 PM UTC

> Once a model is built, the biggest cost is inference

Something i cant find any reliable data for, but would help for a sense of scale: How much use before its equal to training?

I.e. assuming you have the training data and setup, and we only care for compute - How hours of using eg Kimi K3 / Fable, for it to equal the compute required to train it?

Re: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravelling

#120

Post by LarsDu88 » Mon, Jul 20, 2026, 5:11 PM UTC

searealist wrote: post #92

LarsDu88 wrote: post #48

The open weight, open architecture releases of the past several days has me more convinced that ultimately, the winner will be whoever burns their models to ASICs fastest. The LLMs themselves are capable of doing some aspects of chip design as evinced by the K3 press release. Furthermore, the frontier models are "good enough" for a wide swathe of tasks and will soon hit that threshold for a good amount of software en…

How does an ASIC manage to have 60x the memory bandwidth needed to achieve that speedup?

Currently the setup is paged view in RAM shuttled to HBDRAM (VRAM) on the GPU, which in turn has to get materialized piece by piece onto cache SRAM on the GPU.

Cerebras tries to get around this by keeping everything on cache SRAM as much as possible, which it burns directly to the chip wafer itself and physically places that SRAM directly next to the tiny compute unit that does the actual math.

An ideal setup (not sure how easy this is to achieve in practice), is the burn the weights of the model directly to the chip as a sort of ROM, the actual math operations as actual digital circuits, and have SRAM, or even something akin to naked registers to directly compute off inference batch data. Cuts out 2-3 layers of abstraction and indirection.