Download Database Bundles
Want to work with our data locally? Download our version-controlled database bundles below and explore FoodAtlas data on your machine
Our extensive food composition database contains only evidence-based data that can be traced back to its source. As a USDA-NSF funded research project, the data is presented as a free resource under the Apache-2.0 license.
How to Cite
If you use FoodAtlas in your research, please cite:
Li, F., Youn, J., Xie, K., Chan, T., Gupta, P., Yoo, A., ... & Tagkopoulos, I. (2026). A unified knowledge graph linking foodomics to chemical-disease networks and flavor profiles. npj Science of Food. https://doi.org/10.1038/s41538-025-00680-9
| version | release date | summary | size | |
|---|---|---|---|---|
v4.8 | 2026-07-24 | Weekly update — 46 new associations (46 new food–chemical associations across 15 foods and 20 chemicals) from 25 new papers.
Extraction quality (LLM-as-a-judge, 100 sample sentences): precision 1.00, recall 0.73, F1 0.84. | 53.6 MB | |
v4.7 | 2026-07-17 | Weekly update — 69 new associations (68 new food–chemical associations across 31 foods and 42 chemicals) from 35 new papers.
Extraction quality (LLM-as-a-judge, 100 sample sentences): precision 1.00, recall 0.71, F1 0.83. | 53.5 MB | |
v4.6 | 2026-07-10 | Weekly update — 64 new associations (64 new food–chemical associations across 25 foods and 33 chemicals) from 22 new papers.
Extraction quality (LLM-as-a-judge, 100 sample sentences): precision 1.00, recall 0.75, F1 0.86. | 53.5 MB | |
v4.5 | 2026-07-07 | Weekly update — 225 new food–chemical associations across 76 foods and 108 chemicals from 95 new papers.
Extraction quality (LLM-as-a-judge, 100 sample sentences): precision 1.00, recall 0.72, F1 0.84. | 53.5 MB | |
v4.4 | 2026-06-27 | Weekly update — 78 new associations (78 new food–chemical associations across 32 foods and 48 chemicals) from 42 new papers. | 53.8 MB | |
v4.3 | 2026-06-19 | Weekly update — 136 new associations (107 new food–chemical associations across 40 foods and 56 chemicals) from 54 new papers. | 53.8 MB | |
v4.2 | 2026-06-12 | Weekly update — 359 new associations (308 new food–chemical associations across 52 foods and 152 chemicals) from 65 new papers.
Extraction quality (LLM-as-a-judge, 100 sample sentences): precision 0.98, recall 0.69, F1 0.81. | 53.8 MB | |
v4.1 | 2026-06-05 | FoodAtlas v4.1 updates the knowledge graph's literature-extracted food–chemical associations by reprocessing our existing corpus with Claude Opus 4.8, Anthropic's most capable model as of June 2026, improving extraction quality across 124,488 attestations. Going forward, new literature will be processed using Gemma 4 (gemma-4-31B-it), enabling continuous KG updates at scale. | 53.7 MB | |
v4.0 | 2026-04-21 | Redesigned KG schema and entity-mapping pipeline. Added evidence from the Dairy Molecule Database, expanded CTD coverage, and literature extracted with GPT-5.2. Retired previous releases. | 68.1 MB |
Versions prior to v4.0 are retired and no longer available for download.
About AIFS
The AI Institute for Next Generation Food Systems, or AIFS aims to meet growing demands in our food supply by increasing efficiencies using Al and bioinformatics spanning the entire system–from growing crops through consumption. We are dedicated to creating AI applications for a healthier, more sustainable planet from farm to fork.
Connect with us
Subscribe to our newsletter to stay up-to-date on AIFS events, industry news, and AI research.