Cerebras Inference

“It caps output at 32K tokens so the 128K reasoning (the author budget) didn't fit. Cerebras speed is real though”

“Even with thinking off on Cerebras, a problem takes roughly 3 minutes and the 100 problems about 1.5 hours.”

“they aren't reliable at all, because they have to share limited resources dedicated to sales”

“they aggressively rate limit their APIs to the point where if you just run a single coding agent, your coding session will constantly stall out”

“the speed is amazing and I can see the potential with having almost instant inference, but the models are so outdated at this point”

“they have more demand and commitments than their production and scaling capacity”

“Completely unusable last time I tried it. You got hit with rate limits after the first few minutes of using it.”

“1500 tok/s is an impressive speed, and that speed is very important for agent tasks”

“Their architecture recomputes every time so at 150k context every request will have to spend 1.5 min waiting”

“Cerebras is super constrained on capacity right now, all the support is going to enterprise customers.”

“They do offer API services to individual users... though with a set of models that makes it unlikely that you want to use it.”

“Cerebras running Sol at 750 tokens per second, the harness "speed" becomes irrelevant”

“I like Cerabras, but I really wish they would make more of their hosted models generally available.”

“They're pretty supply constrained right now though and their production costs seem prohibitive.”

“using Cerebras at ~2k tokens/s, with very low latency is like a vision into the future”

“Cerebras currently does not provide any discounts for prefix caching making its use for agentic workloads sqr(n_turns) more expensive.”

“watching it slam out several entire screens of thinking/reasoning per second is amazing”

“The best work in this area is memory-integrated Big-Ass-Die or Big-Ass-Chiplet solutions like Cerebras”

“Cerebras are only serving kimi for dedicated endpoint customers; for that you need a >$5m annual deal with them”

“Much harder to catch issues when it is moving a million miles an hour and then it is too late and it has already spent a ton of tokens.”

“Can somebody explain why there’s so little apparent progress here despite the theoretically massive advantage?”

“Cerebras running Qwen 3 235B Instruct at 1.4k tps for cheaper than Claude Haiku”

“the technology is very promising just like Cerebras, but we need to see whether they are able to keep up”

“We already have Groq, Celebras and AWS Bedrock and others in the inference of open models space”

“the $50/month or pay as you go was just not reliable. It would give service unavailable errors or falsely claim we were over our rate limit.”

“I use them heavily in production because OSS 120B on Groq/Cerebras is one of the fastest models available”

“It’s incredible to run a strong model at interactive latencies for whole results.”

“use cerebras with claude code and watch it generate millions of tokens in a few minutes”

“I still enjoy using GLM 4.7 on Cerebras because of the speed you can get there”

“Vertex, Groq, Cerebras, etc. all do not rely on GPUs for much or all of their inference and are all the better for it.”

“seeing two pages of written content output in the literal blink of an eye is magical”

“Cerebras, who get silly fast token generation, which enables interesting cases”

“the ability to run them at 1k tokens per second on Cerebras would be a massive benefit”

“And while the Cerebras service is pretty okay, their website otherwise kinda sucks - and yet you can find clear info!”

“most fixed cloud subscriptions will run out really quickly and will be insufficient (Cerebras being an exception)”

“I love Cerebras. I also love that they've started to scale rate limits to useful levels”

“Cheap enough for now, but of all the companies selling inference at a loss, Cerebras and Groq are probably losing the most per token”

“24M tokens per day for 50 USD seems like good value even with 128k context limitation”

“With Cerebras it's not even worth switching tabs, because it'll finish the same task in like three seconds.”

“They are _really_ fast, and they might make the whole thing actually viable in terms of speed.”