Latest news with #AndrewFeldman
Forbes
08-07-2025
- Business
- Forbes
Who Needs Big AI Models?
Cerebras Systems CEO and Founder Andrew Feldman The AI world continues to evolve rapidly, especially since the introduction of DeepSeek and its followers. Many have concluded that enterprises don't really need the large, expensive AI models touted by OpenAI, Meta, and Google, and are focusing instead on smaller models, such as DeepSeek V2-Lite with 2.4B parameters, or Llama 4 Scout and Maverick with 17B parameters, which can provide decent accuracy at a lower cost. It turns out that this is not the case for coders, or more accurately, the models that can and will replace many coders. Nor does the smaller-is-better mantra apply to reasoning or agentic AI, the next big thing. AI code generators require large models that can handle a wider context window, capable of accommodating approximately 100,000 lines of code. Mixture of expert (MOE) models supporting agentic and reasoning AI is also large. But these massive models are typically quite expensive, costing around $10 to $15 per million output tokens on modern GPUs. Therein lies an opportunity for novel AI architectures to encroach on GPUs' territory. Cerebras Systems Launches Big AI with Qwen3-235B Cerebras Systems (a client of Cambrian-AI Research) has announced support for the large Qwen3-235B, supporting 131K context length (about 200–300 pages of text), four times what was previously available. At the RAISE Summit in Paris, Cerebras touted Alibaba's Qwen3-235B, which uses a highly efficient mixture-of-experts architecture to deliver exceptional compute efficiency. But the real news is that Cerebras can run the model at only $0.60 per million input tokens and per million output tokens—less than one-tenth the cost of comparable closed-source models. While many consider the Cerebras wafer-scale engine expensive, this data turns that perception on its head. Agents are a use case that frequently requires very large models. One question I frequently get is, if Cerebras is so fast, why don't they have more customers? One reason is that they have not supported large context windows and larger models. Those seeking to develop code, for example, do not want to break down the problem into smaller fragments to fit, say, a 32KB context. Now, that barrier to sales has evaporated. 'We're seeing huge demand from developers for frontier models with long context, especially for code generation,' said Cerebras Systems CEO and Founder Andrew Feldman. "Qwen3-235B on Cerebras is our first model that stands toe-to-toe with frontier models like Claude 4 and DeepSeek R1. And with full 131K context, developers can now use Cerebras on production-grade coding applications and get answers back in less than a second instead of waiting for minutes on GPUs.' Cerebras is not just 30 times faster, it is 92% cheaper than GPUs. Cerebras has quadrupled its context length support from 32K to 131K tokens—the maximum supported by Qwen3-235B. This expansion directly impacts the model's ability to reason over large codebases and complex documentation. While 32K context is sufficient for simple code generation use cases, 131K context enables the model to process dozens of files and tens of thousands of lines of code simultaneously, allowing for production-grade application development. Cerebras is 15-100 times more affordable than GPUs when running Qwen3-235B Qwen3-235B excels at tasks requiring deep logical reasoning, advanced mathematics, and code generation, thanks to its ability to switch between "thinking mode" (for high-complexity tasks) and "non-thinking mode" (for efficient, general-purpose dialogue). The 131K context length allows the model to ingest and reason over large codebases (tens of thousands of lines), supporting tasks such as code refactoring, documentation, and bug detection. Cerebras also announced the further expansion of its ecosystem, with support from Amazon AWS, as well as DataRobot, Docker, Cline, and Notion. The addition of AWS is huge; Cerebras has added AWS to its cloud portfolio. Where is this heading? Big AI has constantly been downsized and optimized, with orders of magnitude of performance gains, model sizes, and price reductions. This trend will undoubtedly continue, but will be constantly offset by increases in capabilities, accuracy, intelligence, and entirely new features across modalities. So, if you want last year's AI, you're in great shape, as it continues to get cheaper. But if you want the latest features and functions, you will require the largest models and the longest input context length. It's the Yin and Yang of AI.
Business Wire
28-05-2025
- Business
- Business Wire
Cerebras Beats NVIDIA Blackwell in Llama 4 Maverick Inference
SUNNYVALE, Calif.--(BUSINESS WIRE)--Last week, Nvidia announced that 8 Blackwell GPUs in a DGX B200 could demonstrate 1,000 tokens per second (TPS) per user on Meta's Llama 4 Maverick. Today, the same independent benchmark firm Artificial Analysis measured Cerebras at more than 2,500 TPS/user, more than doubling the performance of Nvidia's flagship solution. 'Cerebras has beaten the Llama 4 Maverick inference speed record set by NVIDIA last week. Artificial Analysis benchmarked Cerebras' Llama 4 Maverick endpoint at 2,522 t/s compared to NVIDIA Blackwell's 1,038 t/s for the same model." - Artificial Analysis Share 'Cerebras has beaten the Llama 4 Maverick inference speed record set by NVIDIA last week,' said Micah Hill-Smith, Co-Founder and CEO of Artificial Analysis. 'Artificial Analysis has benchmarked Cerebras' Llama 4 Maverick endpoint at 2,522 tokens per second, compared to NVIDIA Blackwell's 1,038 tokens per second for the same model. We've tested dozens of vendors, and Cerebras is the only inference solution that outperforms Blackwell for Meta's flagship model.' With today's results, Cerebras has set a world record for LLM inference speed on the 400B parameter Llama 4 Maverick model, the largest and most powerful in the Llama 4 family. Artificial Analysis tested multiple other vendors, and the results were as follows: SambaNova 794 t/s, Amazon 290 t/s, Groq 549 t/s, Google 125 t/s, and Microsoft Azure 54 t/s. Andrew Feldman, CEO of Cerebras Systems, said, 'The most important AI applications being deployed in enterprise today—agents, code generation, and complex reasoning—are bottlenecked by inference latency. These use cases often involve multi-step chains of thought or large-scale retrieval and planning, with generation speeds as low as 100 tokens per second on GPUs, causing wait times of minutes and making production deployment impractical. Cerebras has led the charge in redefining inference performance across models like Llama, DeepSeek, and Qwen, regularly delivering over 2,500 TPS/user.' With its world record performance, Cerebras is the optimal solution for Llama 4 in any deployment scenario. Not only is Cerebras Inference the first and only API to break the 2,500 TPS/user milestone on this model, but unlike the Nvidia Blackwell used in the Artificial Analysis benchmark, the Cerebras hardware and API are available now. Nvidia used custom software optimizations that are not available to most users. Interestingly, none of the Nvidia's inference providers offer a service at Nvidia's published performance. This suggests that in order to achieve 1000 TPS/user, Nvidia was forced to reduce throughput by going to batch size 1 or 2, leaving the GPUs at less than 1% utilization. Cerebras, on the other hand, achieved this record-breaking performance without any special kernel optimizations, and it will be available to everyone through Meta's API service coming soon. For cutting-edge AI applications such as reasoning, voice, and agentic workflows, speed is paramount. These AI applications gain intelligence by processing more tokens during the inference process. This can also make them slow and force customers to wait. And when customers are forced to wait, they leave and go to competitors who provide answers faster—a finding Google showed with search more than a decade ago. With record-breaking performance, Cerebras hardware and resulting API service is the best choice for developers and enterprise AI users around the world. For more information, please visit
Yahoo
28-05-2025
- Business
- Yahoo
Cerebras Beats NVIDIA Blackwell in Llama 4 Maverick Inference
Cerebras Breaks the 2,500 Tokens Per Second Barrier with Llama 4 Maverick 400B SUNNYVALE, Calif., May 28, 2025--(BUSINESS WIRE)--Last week, Nvidia announced that 8 Blackwell GPUs in a DGX B200 could demonstrate 1,000 tokens per second (TPS) per user on Meta's Llama 4 Maverick. Today, the same independent benchmark firm Artificial Analysis measured Cerebras at more than 2,500 TPS/user, more than doubling the performance of Nvidia's flagship solution. "Cerebras has beaten the Llama 4 Maverick inference speed record set by NVIDIA last week," said Micah Hill-Smith, Co-Founder and CEO of Artificial Analysis. "Artificial Analysis has benchmarked Cerebras' Llama 4 Maverick endpoint at 2,522 tokens per second, compared to NVIDIA Blackwell's 1,038 tokens per second for the same model. We've tested dozens of vendors, and Cerebras is the only inference solution that outperforms Blackwell for Meta's flagship model." With today's results, Cerebras has set a world record for LLM inference speed on the 400B parameter Llama 4 Maverick model, the largest and most powerful in the Llama 4 family. Artificial Analysis tested multiple other vendors, and the results were as follows: SambaNova 794 t/s, Amazon 290 t/s, Groq 549 t/s, Google 125 t/s, and Microsoft Azure 54 t/s. Andrew Feldman, CEO of Cerebras Systems, said, "The most important AI applications being deployed in enterprise today—agents, code generation, and complex reasoning—are bottlenecked by inference latency. These use cases often involve multi-step chains of thought or large-scale retrieval and planning, with generation speeds as low as 100 tokens per second on GPUs, causing wait times of minutes and making production deployment impractical. Cerebras has led the charge in redefining inference performance across models like Llama, DeepSeek, and Qwen, regularly delivering over 2,500 TPS/user." With its world record performance, Cerebras is the optimal solution for Llama 4 in any deployment scenario. Not only is Cerebras Inference the first and only API to break the 2,500 TPS/user milestone on this model, but unlike the Nvidia Blackwell used in the Artificial Analysis benchmark, the Cerebras hardware and API are available now. Nvidia used custom software optimizations that are not available to most users. Interestingly, none of the Nvidia's inference providers offer a service at Nvidia's published performance. This suggests that in order to achieve 1000 TPS/user, Nvidia was forced to reduce throughput by going to batch size 1 or 2, leaving the GPUs at less than 1% utilization. Cerebras, on the other hand, achieved this record-breaking performance without any special kernel optimizations, and it will be available to everyone through Meta's API service coming soon. For cutting-edge AI applications such as reasoning, voice, and agentic workflows, speed is paramount. These AI applications gain intelligence by processing more tokens during the inference process. This can also make them slow and force customers to wait. And when customers are forced to wait, they leave and go to competitors who provide answers faster—a finding Google showed with search more than a decade ago. With record-breaking performance, Cerebras hardware and resulting API service is the best choice for developers and enterprise AI users around the world. For more information, please visit View source version on Contacts pr@
CNBC
16-05-2025
- Business
- CNBC
Cerebras CEO says chipmaker's 'aspiration' is to hold IPO in 2025
Cerebras CEO Andrew Feldman said his hope is to take his company public in 2025 now that the chipmaker has obtained clearance from the U.S. government to sell shares to an entity in the United Arab Emirates. "That's our aspiration," Feldman told reporters on Thursday at the company's Supernova conference in San Francisco, after being asked if an IPO was likely this year. Cerebras, which makes processors for artificial intelligence workloads, filed to go public in September but hasn't provided an update on the expected size or timing of an offering. In March, the company said it had obtained clearance from a U.S. committee to sell shares to Group 42, a Microsoft-backed AI company based in the UAE. That clearance came from the Committee on Foreign Investment in the United States, or CFIUS, and marked a key step for Cerebras in its effort to go public. Cerebras competes with Nvidia, whose graphics processing units (GPUs) are the industry's choice for training and running AI models. More than 85% of Cerebras' revenue in the first half of 2024 came from Group 42. The tech IPO market broadly has been in a drought since early 2022, when rising inflation and higher interest rates pushed investors out of risky assets. Cerebras appeared poised to be the first notable pure-play AI IPO after its filing, but then the came the delay. CoreWeave, which provides AI infrastructure, debuted in March and has seen its market value jump about 65% since its IPO. The IPO market is showing signs of life, with trading app eToro hitting the Nasdaq this week and digital health provider Hinge Health scheduled to go out next week. The Middle East is becoming a more critical market for AI development. Nvidia CEO Jensen Huang was in Riyadh, Saudi Arabia this week along with other tech leaders and President Donald Trump for the Saudi-U.S. Investment Forum. Nvidia said at the event that it will sell more than 18,000 of its latest AI chips to Saudi company Humain. Group 42 is also reportedly on tap to purchase 100,000 GPUs a year as part of a bigger agreement between the U.S. and UAE. Feldman said at the roundtable with reporters that it's "important to be among the big dogs" and said, regarding the latest announcements, "You've got half the story. I can't share the other half." In addition to Microsoft, Cerebras sells to Meta and IBM. Feldman said last year that the company would have another "hyperscaler" within the first half of 2025. "We're close with another," he said on Thursday. "I think they haven't been the quickest to respond." Earlier in the day, Cerebras announced the ability to run an open-source model from Alibaba on its chips at what it says is a lower price than what OpenAI's GPT-4.1 model charges, and at a higher speed.
Yahoo
15-05-2025
- Business
- Yahoo
Cerebras Launches Qwen3-32B: Real-Time Reasoning with One of the World's Most Powerful Open Models
SUNNYVALE, Calif., May 15, 2025--(BUSINESS WIRE)--Cerebras today announced the launch of Qwen3-32B, one of the most advanced open-weight models in the world, now available on the Cerebras Inference Platform. Developed by Alibaba, Qwen3-32B rivals the performance of leading closed models like GPT-4.1 and DeepSeek R1—and now, for the first time, it runs on Cerebras with real-time responsiveness. Qwen3-32B on Cerebras performs sophisticated reasoning and returns the answer in just 1.2 seconds — up to 60x faster than comparable reasoning models such as DeepSeek R1 and OpenAI o3. This is the first reasoning model on any hardware to achieve real-time reasoning. Qwen3-32B on Cerebras is the fastest reasoning model API in the world, ready to power production-grade agents, copilots, and automation workloads. "This is the first time a world-class reasoning model—on par with DeepSeek R1 and OpenAI's o-series — can return answers instantly," said Andrew Feldman, CEO and co-founder of Cerebras. "It's not just fast for a big model. It's fast enough to reshape how real-time AI gets built." The First Real-Time Reasoning Model Reasoning models are widely recognized as the most powerful class of large language models—capable of multi-step logic, tool use, and structured decision-making. But until now, they've come with a tradeoff: latency. Inference often takes 30–90 seconds, making them impractical for responsive user experiences. Cerebras eliminates that bottleneck. Qwen3-32B delivers first-token latency in just one second, and completes full reasoning chains in real time. This is the only solution on the market today that combines high intelligence with real-time speed—and it's available now. Transparent, Scalable Pricing Qwen3-32B is available on Cerebras with simple, production-ready pricing: $0.40 per million input tokens $0.80 per million output tokens This is 10x cheaper than GPT-4.1, while offering comparable or better performance. All developers receive 1 million free tokens per day, with no waitlist. Qwen3-32B is fully open-weight and Apache 2.0 licensed, and can be integrated in seconds using standard OpenAI- or Claude-compatible endpoints. Qwen3-32B is live now on For teams seeking to build fast, intelligent, production-ready AI systems, it is the most powerful open model you can use today. About Cerebras Systems Cerebras Systems is a team of pioneering computer architects, computer scientists, deep learning researchers, and engineers of all types. We have come together to accelerate generative AI by building from the ground up a new class of AI supercomputer. Our flagship product, the CS-3 system, is powered by the world's largest and fastest commercially available AI processor, our Wafer-Scale Engine-3. CS-3s are quickly and easily clustered together to make the largest AI supercomputers in the world, and make placing models on the supercomputers dead simple by avoiding the complexity of distributed computing. Cerebras Inference delivers breakthrough inference speeds, empowering customers to create cutting-edge AI applications. Leading corporations, research institutions, and governments use Cerebras solutions for the development of pathbreaking proprietary models, and to train open-source models with millions of downloads. Cerebras solutions are available through the Cerebras Cloud and on premise. For further information, visit or follow us on LinkedIn or X. View source version on Contacts Media Contact Press Contact: PR@ Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
