AI glossary Responsibility
What is a model card in AI?
← All glossary termsA short, structured document that describes an AI model's intended use, training data, evaluation results, limitations and ethical considerations.
Explained
A model card is documentation that travels with a trained model and explains what it is for and how well it works. The idea was proposed by researchers at Google in 2018: a standard format to report a model's intended use, the data it was trained and evaluated on, its performance across relevant groups and conditions, and its known limitations and risks. Model cards help people decide whether a model fits their use case, and they support governance, procurement and audit readiness alongside technical specifications. Model hubs such as Hugging Face now attach a model card to every published model.
What a model card contains
- Model details: who built it, version, type of model and licence.
- Intended use: the uses it was designed for, and uses that are out of scope.
- Training and evaluation data: sources, coverage and known gaps.
- Metrics and results: how it was evaluated and how it performs, ideally broken down by relevant groups and conditions.
- Limitations, risks and ethical considerations: where it fails, biases found and recommendations for safe use.
Why model cards matter for enterprises
When a company uses a model built by someone else, the model card is often the first evidence of what it can and cannot do. It helps procurement and risk teams compare options, helps engineers decide what to test, and gives auditors a record of what was known when the model was chosen.
For models a company builds itself, writing a model card forces the team to state the intended use and the evaluation openly. It is also a practical starting point for the technical documentation that regulation such as the EU AI Act requires for some systems.
Model cards, system cards and datasheets
A model card documents one model. A system card documents a whole AI system or product, including safeguards and how the model is deployed. A datasheet for a dataset documents the data: how it was collected, what it contains and how it should be used. Together they give a fuller picture than any one of them alone.
Frequently asked questions
Who invented model cards?
Model cards were proposed in the paper 'Model Cards for Model Reporting' by Margaret Mitchell and colleagues at Google, first published in 2018 and presented at the FAT* conference in 2019.
What should a model card include?
At least the model's details, intended and out-of-scope uses, training and evaluation data, performance metrics, and limitations, risks and ethical considerations. Breaking results down by relevant groups makes it far more useful.
What is a model card on Hugging Face?
On the Hugging Face Hub, the model card is the README of a model repository: a Markdown file with metadata that describes the model, its uses, training, evaluation and limitations.
What is the difference between a model card and a system card?
A model card describes a single model. A system card describes a complete AI system, including how models are combined, the safeguards around them and how they behave in deployment.
Are model cards required by the EU AI Act?
The AI Act does not use the term model card, but it requires technical documentation and information for deployers for high-risk systems and documentation duties for general-purpose AI models. A good model card covers much of that content.
Decision guides that use this term
How to put it to work
Where it shows up at Thinkia
Further reading
Related terms
Sources
- Mitchell et al. · Model Cards for Model Reporting (arXiv)
- Hugging Face · Model cards documentation
- TensorFlow · Model Card Toolkit (Responsible AI guide)
Updated: 2026-10-21