God I wish they'd back up all of those claims by offering a subscription of Kimi K3 and GLM 5.3, not some outdated GLM 4.7 instance that they then proceed to call a preview model and say that they'll remove it, leaving users only with GPT-OSS 120B which is nigh useless nowadays: https://support.cerebras.net/articles/9996007307-cerebras-co... and https://www.cerebras.ai/pricing
Guess they don't care about regular devs atm and are focused only on hardware sales.
show comments
xmorse
Cerebras is very fast but you can basically never use it because of its scarcity
syntaxing
I think the fun takeaway from this is that GPT 5.4 is probably 45B active parameters and GPT 5.6 Sol is closer to 50B.
show comments
sreekanth850
AMD along with cerebras may probably compete with NVIDIA monopoly in near future. Also, NVIDIA will have competition form multiple companies. Just my prediction.
show comments
rajnathani
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.
reilly3000
> 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?
show comments
ethanzhang1024
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?
show comments
aneryu
It would be even better if a version available to individual users were released soon.
show comments
anonymous_user9
Conspicuously missing: power consumption figures
show comments
9cb14c1ec0
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!
show comments
kobe_bryant
can these vibe coded sites please set a max width and overflow so their sites work fine on mobile
show comments
denizay
The comparison seems incomplete. CS‑4 is a full rack-scale system with three wafer-scale processors, but the exact GPU models, GPU count, power consumption, price information are not disclosed.
We still don't know if buying a multi-GPU rack (or racks) is cheaper and/or more efficient in power.
The fact that they didn't disclose these numbers makes me believe that the numbers are not in their favor. And personally, makes me see them as disingenuous.
lostmsu
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?
show comments
selimonder
That "GPU" comparison is the vaguest i seen so far
show comments
arthurcolle
What's the sticker price? If I have 20 million in the bank can I just like buy one or what
show comments
sva_
> enabling massive clusters and models with more than 50 trillion parameters
4k0hz
> Introducing the all new Cerebras CS-4, a revolutionary rack-scale solution that delivers upto 30x faster inference compared to GPUs, enhanced economics, and a simple path todeploy [sic] hyperscale capacity.
Did nobody proofread this?
show comments
avantnyc
Cerebras should slowly also move to dgx/ryzen market for a desktop version for masses at affordable price yet providing substantial tokens/second on desktop
show comments
tamimio
I wonder what are the benchmarks of hashcat on different hashes.
OutOfHere
Five years from now, I don't know why anyone will still be using Nvidia for inference. Note that Cerebras is for inference only, not for training.
I understand that Cerebras has competition, but this bodes even more poorly for Nvidia for inference. Nvidia may still have a role to play for training, however.
show comments
adventured
OpenAI needs to immediately move to acquire Cerebras.
Nvidia's extreme margin is the opportunity for OpenAI's cost reduction. Buying Cerebras would pay for itself and they should take all of its future production (after filling required contracts).
Right now China's models have no silicon moat. Cerebras as a drastic speed-up / cost-reduction potential, can assist in building a competitive moat. And every time a Cerebras pops up, OpenAI or Anthropic should eat them if at all possible.
There's no stand-alone frontier AI company of great scale in the near future that doesn't have a large silicon advantage in-house. Apple knew it in smartphones, Google figured it out a long time ago as well.
show comments
gpm
Is it just me or is it bizarre that they're advertising old open-weight models.
GLM 4.7 (December 2025) not 5 (Feb) 5.1 (April) or 5.2 (June). 5.3 (4 days ago) is, to be fair, not open weights yet... but there's a lot since 4.7.
Kimi K2.7 (April) not K2.7-code (June) or K3 (July).
Gemma 4 (April), Llama (April), and gpt-oss (August 2025) are up to date, but old (for models).
Meanwhile the closed source GPT 5.6 sol is up to date (June)...
Should potential purchasers take away from this that they're not going to be able to run recent models unless they front the cost of developing software or something?
God I wish they'd back up all of those claims by offering a subscription of Kimi K3 and GLM 5.3, not some outdated GLM 4.7 instance that they then proceed to call a preview model and say that they'll remove it, leaving users only with GPT-OSS 120B which is nigh useless nowadays: https://support.cerebras.net/articles/9996007307-cerebras-co... and https://www.cerebras.ai/pricing
Guess they don't care about regular devs atm and are focused only on hardware sales.
Cerebras is very fast but you can basically never use it because of its scarcity
I think the fun takeaway from this is that GPT 5.4 is probably 45B active parameters and GPT 5.6 Sol is closer to 50B.
AMD along with cerebras may probably compete with NVIDIA monopoly in near future. Also, NVIDIA will have competition form multiple companies. Just my prediction.
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.
> 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?
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?
It would be even better if a version available to individual users were released soon.
Conspicuously missing: power consumption figures
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!
can these vibe coded sites please set a max width and overflow so their sites work fine on mobile
The comparison seems incomplete. CS‑4 is a full rack-scale system with three wafer-scale processors, but the exact GPU models, GPU count, power consumption, price information are not disclosed. We still don't know if buying a multi-GPU rack (or racks) is cheaper and/or more efficient in power. The fact that they didn't disclose these numbers makes me believe that the numbers are not in their favor. And personally, makes me see them as disingenuous.
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?
That "GPU" comparison is the vaguest i seen so far
What's the sticker price? If I have 20 million in the bank can I just like buy one or what
> enabling massive clusters and models with more than 50 trillion parameters
> Introducing the all new Cerebras CS-4, a revolutionary rack-scale solution that delivers upto 30x faster inference compared to GPUs, enhanced economics, and a simple path todeploy [sic] hyperscale capacity.
Did nobody proofread this?
Cerebras should slowly also move to dgx/ryzen market for a desktop version for masses at affordable price yet providing substantial tokens/second on desktop
I wonder what are the benchmarks of hashcat on different hashes.
Five years from now, I don't know why anyone will still be using Nvidia for inference. Note that Cerebras is for inference only, not for training.
I understand that Cerebras has competition, but this bodes even more poorly for Nvidia for inference. Nvidia may still have a role to play for training, however.
OpenAI needs to immediately move to acquire Cerebras.
Nvidia's extreme margin is the opportunity for OpenAI's cost reduction. Buying Cerebras would pay for itself and they should take all of its future production (after filling required contracts).
Right now China's models have no silicon moat. Cerebras as a drastic speed-up / cost-reduction potential, can assist in building a competitive moat. And every time a Cerebras pops up, OpenAI or Anthropic should eat them if at all possible.
There's no stand-alone frontier AI company of great scale in the near future that doesn't have a large silicon advantage in-house. Apple knew it in smartphones, Google figured it out a long time ago as well.
Is it just me or is it bizarre that they're advertising old open-weight models.
GLM 4.7 (December 2025) not 5 (Feb) 5.1 (April) or 5.2 (June). 5.3 (4 days ago) is, to be fair, not open weights yet... but there's a lot since 4.7.
Kimi K2.7 (April) not K2.7-code (June) or K3 (July).
Gemma 4 (April), Llama (April), and gpt-oss (August 2025) are up to date, but old (for models).
Meanwhile the closed source GPT 5.6 sol is up to date (June)...
Should potential purchasers take away from this that they're not going to be able to run recent models unless they front the cost of developing software or something?