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

University and research, United States

Berkeley

UC Berkeley's labs gave the world Vicuna, Chatbot Arena and vLLM, a common engine for serving models.

The University of California, Berkeley, across the bay from San Francisco, is one of the world's leading centres for machine-learning research. Its Berkeley AI Research lab and Sky Computing Lab have produced widely used open tools for building, serving and comparing language models, and many founders and researchers at the nearby labs trained here.

Vicuna and the Arena

In March 2023 LMSYS, a group of researchers from Berkeley, Stanford, Carnegie Mellon, UC San Diego and MBZUAI, released Vicuna, a Llama model fine-tuned on conversations that users had shared from ChatGPT. To compare such chatbots, the team launched Chatbot Arena in May 2023, where visitors vote between two anonymous answers. Its leaderboard became one of the most watched in the field and grew into LMArena.

Serving models

Berkeley researchers created vLLM, an open-source engine for running language models, introduced with the PagedAttention method in 2023. PagedAttention manages the memory a model uses to remember earlier tokens much as an operating system manages pages of memory. vLLM is now used by many companies and clouds. Berkeley also produced the Gorilla project on models that call software tools.

Models of Berkeley

The weekly digest

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