AllNutrition API
An evidence-based nutrition API. Every answer is grounded in peer-reviewed research, clinical guidelines, and expert position statements curated by AllNutrition, and ships with citations to the underlying sources.
What you can build
- Cited nutrition assistants — ground GPT, Claude, or Gemini-powered chat in real research instead of hallucinations.
- Health and fitness apps — surface evidence-graded dietary recommendations.
- Internal tooling — equip dietitians and clinicians with a search-friendly research library.
- Custom GPTs & tool-using agents — drop the OpenAPI spec straight into a GPT Action or LangChain tool.
The endpoints
The public API lives at https://www.allnutrition.info.
Ask a nutrition question. Returns a Markdown answer plus a list of citations, each tagged with an evidence tier and trust score.
Semantic search over the curated reference library — no LLM generation. Use when you want to bring your own model or surface citations directly in a UI.
Liveness probe. Not rate limited.
Why citations matter here
Generic LLMs frequently invent facts about nutrition — dosages, populations, contraindications. AllNutrition is built so that every answer maps to a real, retrievable source. Each response includes:
- An
evidence_strengthlabel (strong | moderate | limited | insufficient) — how solid the underlying evidence is. - A
consensus_levellabel (high | moderate | mixed | low) — how well the cited sources agree. - A list of
sources, each with title, publisher, evidence tier, trust score, publication date, and the exact excerpt used as RAG context.
See Evidence & consensus for the full taxonomy.