News
ReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIFrontierGoogle is making it easier to opt out of Gemini AI model training – here’s howFrontierOpenAI Unveils New Astra AI Model After Scrapping Prior ReleaseReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIFrontierGoogle is making it easier to opt out of Gemini AI model training – here’s howFrontierOpenAI Unveils New Astra AI Model After Scrapping Prior Release
EveryLLM

Lab city, United States

Santa Clara

Nvidia's home town, whose chips train most large models and whose researchers build Nemotron.

Santa Clara, in the heart of Silicon Valley, is the headquarters of Nvidia, whose graphics processors (GPUs) train and run most of the world's large language models. Nvidia is also a model builder: its research teams wrote Megatron-LM, a widely used library for splitting giant models across many chips, and they release the open Nemotron models, often together with their training data.

Chips and software

Nvidia's A100 (2020) and H100 (2022) chips, linked by its NVLink and InfiniBand networking, became the standard hardware for training models such as GPT-4 and Llama 3. Its CUDA software platform, first released in 2007, is what most training code is written for. US export controls limit sales of the most advanced of these chips to China, a policy argued over in Washington and Beijing alike.

Nemotron

Nvidia released Nemotron-4 340B in June 2024 with a licence allowing its outputs to be used to train other models, and has since published many Nemotron models tuned from Llama and Qwen bases as well as its own designs. Nemotron Nano 2 (August 2025) is a hybrid of Mamba-2 and transformer layers. Nearby Sunnyvale is home to Cerebras, whose wafer-scale chips trained the Arabic model Jais.

Models of Santa Clara

The weekly digest

Every Monday: the model of the day, new pages on EveryLLM, and the conferences and anniversaries of the week ahead.