Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- Interestingly they’re still on the WSE-3 (5nm TSMC) wafer chip and slightly bumped up the specs there (overlocking mostly it seems), for why it’s called WSE-3 Turbo now. I think people were also expecting WSE-4, as it’s been 2 years now since WSE-3 was launched.by rajnathani
- Impressive that this is an "interim" product, the start of a new line that ought to be continued with the WSE-4 family, where they are supposed to use a 3nm process and, maybe, 3D stacked SRAM. The modular architecture also points towards field upgrades that are badly needed for AI datacenter builders.by rbanffy
- Conspicuously missing: power consumption figures
- "10x more throughput per watt than CS-3"by logicallee
- 162 kWby wmf
- Cerebras is very fast but you can basically never use it because of its scarcityby xmorse
- If cerebars is performing well, why didn't its predecessor, server S-3, become the largest API token provider on OpenRouter, surpassing the official model releases?
- Because they aren't selling inference, they're selling hardware. The only reason they sell any tokens on OpenRouter is so they get on the benchmark that shows them as the fastest provider. It's free advertising.by 0xbadcafebee
- Cerebras capacity was pretty much entirely bought out at some point. We needed it and couldn't get it.by gampleman
- The WSE is very expensive to build, and they have a waiting list of customers who are already willing to pay a lot of money for the available supply.by aseipp
- If you're willing to pay a significant premium for latency, why use openrouter? And anyway Cerebras only supported a few specific models.by smallerize
- Probably the same reason why there are more people who takes buses, subways, trains than drive Ferraris.by aurareturn
- it only takes ~445 GB300 NVL72 (about $22b) to run ALL of openrouter demand for a year. Microsoft rolled out $32b of DC 2026Q1.
imo the issue is that most openrouter demand is inauthentic activity (things that anthropic and openai models will refuse to do like pretend to not be bots when interacting with humans)
- Cerebras provides high-speed inference at high cost. It's never going to be the cheapest and thus it will probably remain niche.by wmf
- Without having any inside information, one possible theory:
All or a vast majority of of the cerebras manufacturing capacity was going to a few companies that aren't publicly available inference providers on openrouter, for their own internal use.
or
The asking price of the S-3, no matter how speedy it might be, for small/medium size customers made it economically prohibitive to purchase and use to sell public inference vs. buying more common nvidia b200 or whatever.
by walrus01 - KV caching status?
What's the point of 1000tok/s if you have to do prefill on every agentic turn which at 100k depth would make it 1.5 min latency every turn?
by lostmsu - Information about RAM type/size and connection topology of the RAM to be used for context cache seems to be conspicuously absent from the slick looking marketing materials.by walrus01
- Just a reminder for everyone that we are only several years and 3 or 4 iterations into hardware being optimized for LLMs. We should all expect orders of magnitude improvement in speed and/or cost over the next 5 years. Then we can have fun conversations about "unlimited" "intelligence" and about what the price wars and profit margins of consumer AI products are when your average ChatGPT user costs the company $0.10 per month.
> CS-4 delivers more than 1,000 tokens per second on models exceeding 10 trillion parameters
Wow!
by 9cb14c1ec0 - Congratulations! You have just realized that the AI data center build out is a total scam, built on both the insurmountable trillions of debt, and the assumption that only GPUs are all we need to continue scaling.
There exist other AI accelerators (TPUs, ASICs) that perfectly exceed the throughput that LLMs need to scale as well. But the true solution is more software optimizations. There's a tiny handful of them but more needs to be discovered so that we can reduce building hundreds of more data centers as the alternatives mature.
As better software becomes more useful for the alternative AI hardware for developers with LLMs running efficiently you then would have more choices of hardware to run your LLMs on rather than just only GPUs.
by rvz - What LLM-specific hardware improvements should one expect? Seems to me that LLM inference is simple architecturally (matmul et al) so most scaling in hardware should come from general improvements (memory BW, packaging, interconnect, power).by brausepulver
- This is part of why I think the data center build-out is a bubble. We've barely scratched the surface when it comes to hardware optimization. We'll see exponential improvements in energy efficiency and speed over the next decade. Exponential, not linear.
GPUs really aren't that great for AI. They just happen to be the best chips we have in mass production right now for this work load, and it takes time to field new designs. Basically every chip engineer on the planet is working on this right now.
by api - A design that bakes the architecture into silicon would be 10x faster, and imagine a version that does all the multiplication ops using single log-amp addition versus dozens of transistors to cut down the amount of silicon used by 50x. The ceiling for AI optimized hardware is extremely high.
Stack on top of that the fact that diffusion based models like the ones made by Inception Labs are far faster and more efficient than autoregressive LLMs and have an even higher ceiling of optimization (single step path prediction via model distillation versus 50 step denoise is currently an active area for image diffusion)
The human brain is soon neither going to be more powerful nor energy efficient than the stuff we use to run AI.
by LarsDu88 - > Then we can have fun conversations about "unlimited" "intelligence" and about what the price wars and profit margins of consumer AI products are when your average ChatGPT user costs the company $0.10 per month.
We can have that discussion now: sounds like that would kill OpenAI and Anthropic
by dgellow - Hence why taalas was one of the best strategic acquisitions of the year.
I'm honestly baffled they were not acquired by somebody else (sorry AMD).
by moralestapia - And, the software side isn't finished being optimized, either. We've seen with Qwen 3.8 27B and DeepSeek V4 Flash 0731 and GLM 5.3 that quite small models can pack a punch. Intelligence density will improve, efficiency of kernels will improve, efficiency of KV caching and MTP will improve, algorithms for splitting workloads across compute units will improve.
It'll all be as cheap as DeepSeek was before the price hike. And, it'll become more and more realistic to run near-frontier intelligence on personal devices.
by SwellJoe - I was hoping to see Cerebras launch something other than GPT-OSS-120b in production this week, especially with GLM4.7 going away.
If they could launch Qwen 27b or Deepseek Flash that would be amazing.
by bearjaws - > CS-4 delivers more than 1,000 tokens per second on models exceeding 10 trillion parameters
Oops did they just out GPT-5.6 sol’s parameter count?
by reilly3000 - cerebras model are different size then the original modelsby kanwisher
- I mean we kinda know the frontier models are multi trillion parameter models. The only open weights that are close to the frontier are that size tooby whatever1
- Sol is supposed to be 5T according to rumour. The imminent Astra is allegedly 10by sho