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
SaulLM-7B

SaulLM-7B

A 7-billion parameter language model for legal text.

SaulLM-7B is a large language model built for the legal domain. Released in March 2024 by an academic team, it uses the Mistral 7B architecture. It was trained on over 30 billion tokens of English legal text to improve document comprehension and generation.

Share on WhatsApp

  • Built on the Mistral 7B architecture
  • Trained on 30 billion legal tokens
  • Released under the MIT License
  • First LLM designed explicitly for law
  • Uses novel instructional fine-tuning

What it is

SaulLM-7B is a specialised large language model designed for legal tasks. It contains 7 billion parameters and focuses on understanding and generating legal text. The model was introduced in March 2024 by a group of researchers including Pierre Colombo. It aims to provide high proficiency in processing legal documents, serving as a tool for professionals in the field.

How it came about

The model was announced in early March 2024 via a paper on arXiv. The authors describe it as the first large language model explicitly created for the legal domain. The development involved leveraging existing open architectures to create a specialised tool for legal comprehension and generation tasks.

Inside the model

The system builds upon the Mistral 7B architecture, which is a dense decoder transformer. Training used a corpus of more than 30 billion tokens drawn from English legal sources. The team applied a novel instructional fine-tuning method using legal datasets. This process aimed to enhance performance on specific legal tasks beyond general language capabilities.

What it can do

The creators state that the model shows state-of-the-art skill in understanding legal documents. It is tailored for both comprehension and generation within the legal sphere. The instructional fine-tuning helps it handle specific legal queries and text processing tasks more effectively than general purpose models of similar size.

How to use it

The weights are available under the MIT License. This permissive licence allows for broad use and modification. Users can access the model through platforms that host open weights, such as Hugging Face, facilitating integration into other legal tech applications or research projects.

Questions people ask about SaulLM-7B

What is SaulLM-7B used for?

It is designed for legal text comprehension and generation. The model helps users process and understand legal documents with high proficiency.

Is SaulLM-7B free to use?

Yes, the model is released under the MIT License. This permits open use and modification of the weights for various applications.

What data was it trained on?

It was trained on an English legal corpus containing over 30 billion tokens. This specific data helps it specialise in legal language and concepts.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on SaulLM-7B yet, so the facts on this page come from the sources listed above: arxiv.org/abs/2403.03883. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryLLM wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

    The weekly digest

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