Galactica
A large language model for science. Can summarize academic literature, solve math problems, generate Wiki articles, write scientific code, annotate molecules and proteins, and more. [Model API](https://github.com/paperswithcode/galai).
What is Galactica?
Galactica is a large language model trained specifically on scientific text and data. It can read and summarise academic papers, work through mathematical problems, generate scientific writing, produce working code for research tasks, and annotate molecular and protein structures. The tool is designed for researchers, scientists, and students who need to process scientific information quickly or generate technical content. Unlike general-purpose language models, Galactica understands scientific notation, chemical formulas, and domain-specific concepts. You can use it to speed up literature reviews, assist with computational work, or help write up research documentation. It's open source, so you can run it locally or integrate it into your own tools via the API.
Key Features
Academic literature summarisation
Extract key findings from research papers and generate concise summaries
Mathematical problem solving
Work through calculations and mathematical reasoning tasks
Scientific writing generation
Create wiki-style articles, lab notes, and research documentation
Code generation
Write and debug scientific Python code for research applications
Molecular and protein annotation
Analyse and label chemical structures and biological sequences
Open source API
Deploy the model locally or integrate it into custom applications
Pros & Cons
Advantages
- Specialised for science rather than general tasks, so it understands technical concepts better than broad models
- Completely open source and free to use, with no licensing fees or usage limits
- Can be self-hosted, giving you full control over your data and infrastructure
- Useful for automating repetitive research tasks like paper summarisation or code scaffolding
Limitations
- Smaller and less capable than the largest general-purpose language models, so results can be less polished
- Requires some technical knowledge to deploy and integrate via the API
- Performance on non-English scientific content is likely limited
Use Cases
Summarising batches of research papers to identify relevant literature quickly
Generating initial drafts of methods sections or technical documentation
Writing boilerplate code for data analysis or simulation tasks
Learning how to approach scientific problems by working through examples
Building custom research tools that need to understand scientific notation and concepts