“Cerebras is capable of offering Llama 3.1 405B at 969 output tokens/s and they have announced they will soon be offering a public inference endpoint 🏁 We have independently benchmarked a private endpoint shared by @CerebrasSystems and have measured 969 output tokens/s, >10X
[AINews] Llama 3.2: On-device 1B/3B, and Multimodal 11B/90B (with AI2 Molmo kicker) • Buttondown
Llama 3.1 405B now runs at 969 tokens/s on Cerebras Inference – Cerebras
“wow @deepseek_ai R1 cooks. people are sleeping on the competition for AI from China. clearly chip controls not working
Exclusive | Sam Altman seeks backers for AI chipmaker to challenge Nvidia: source
“🌟 Inference Scaling Laws of DeepSeek-R1-Lite-Preview Longer Reasoning, Better Performance. DeepSeek-R1-Lite-Preview shows steady score improvements on AIME as thought length increases.
Amazon
“We’re expanding our collaboration with AWS. This includes a new $4 billion investment from Amazon and establishes AWS as our primary cloud and training partner.
Huawei
Exclusive: Huawei aims to mass-produce newest AI chip in early 2025, despite US curbs | Reuters
NVIDIA
“15 minutes into $NVDA call, key insights so far: • NVIDIA reported record Q3 revenue of $35.1 billion, up 17% sequentially and 94% year-on-year, surpassing their outlook of $32.5 billion. • Data center revenue reached $30.8 billion, marking a 17% sequential and 112%” / X
“🚀 Introducing LLaMA-Mesh! 🎉 We fine-tuned LLaMA on 3D Mesh data, enabling LLMs to natively generate 3D meshes via chatting while retaining their original language capabilities. ✨ Model weights and inference code are fully open-sourced. 🌐 Project page:
LLaMA-Mesh
“Creativity unleashed 👨🎨🎨✨ 👀 Edify3D from #NVIDIAResearch explores a new way to generate text-to-#3D images. 📗 Project page:
Nvidia rivals focus on building a different kind of chip to power AI products | AP News
““We are now generating intelligence at scale.” The remarks from Jensen Harris at tonight’s @ComputerHistory honoring of its 2024 fellows. A different side of Jensen.
“Key insights from $NVDA call: • NVIDIA reported record Q3 revenue of $35.1 billion, up 17% sequentially and 94% year-on-year, driven by strong growth across all market platforms, particularly in data centers. • The company highlighted exceptional demand for its NVIDIA Hopper
Nvidia’s Blackwell chips reportedly face overheating issue ahead of Q3 earnings
Nvidia Says Its Blackwell Chip Is Fine, Nothing to See Here | WIRED
New Nvidia AI chips overheating in servers, the Information reports | Reuters
Nvidia (NVDA) earnings report Q3 2025
Leave a Reply