A general-purpose AI model is, under Article 3(63) of the EU AI Act, an AI model trained on large amounts of data that displays significant generality, can competently perform a wide range of distinct tasks and can be integrated into many downstream systems.
What GPAI model means
Chapter V of the AI Act sets obligations for providers of general-purpose AI models, applying since 2 August 2025. Models placed on the market before that date have until 2 August 2027. Article 53 requires technical documentation, information for downstream providers, a copyright policy and a public summary of training content.
Models with systemic risk carry extra duties under Article 55, including model evaluation, adversarial testing, serious incident reporting and cybersecurity protection. A model is presumed to have high-impact capabilities when its training compute exceeds 10^25 floating point operations. The Commission published the General-Purpose AI Code of Practice in July 2025 as a voluntary way to show compliance.
Most buyers do not contract with model providers directly. The model sits behind a vendor, so the useful questions are which model, from whom, and what documentation the vendor received under Article 53. The answers belong in the vendor's DPA, sub-processor list and product documentation.
What to check in vendor documents
- Which general-purpose models the service uses, and their providers.
- Whether the vendor holds the Article 53 downstream documentation and will share relevant parts.
- Whether the model provider has signed the General-Purpose AI Code of Practice.
- How the vendor tests the model for your use case.
Related terms
- EU AI ActThe EU AI Act is Regulation (EU) 2024/1689, the EU's risk-based law on artificial intelligence. It bans some practices, sets duties for high-risk systems, general-purpose AI models and transparency, and splits obligations between providers and deployers.
- Model provider sub-processorA model provider sub-processor is a company whose AI model a vendor uses to deliver its service and which processes your personal data in doing so. Under GDPR it is a sub-processor that needs authorisation and flow-down terms.
- AI vendor riskAI vendor risk is the risk an organisation faces when it buys a product that relies on AI models: how its data is used for training, which model providers process it, how outputs are controlled and how the vendor meets AI rules.
- NIST AI Risk Management Framework (AI RMF)The NIST AI Risk Management Framework (AI RMF) is a voluntary framework from the US National Institute of Standards and Technology for managing the risks of AI systems, organised into four functions: Govern, Map, Measure and Manage.
Sources
- Regulation (EU) 2024/1689 (AI Act)eur-lex.europa.eu/eli/reg/2024/1689/oj
- European Commission, General-Purpose AI Code of Practicedigital-strategy.ec.europa.eu/en/policies/contents-code-gpai
- European Commission, AI Act regulatory frameworkdigital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
Summarised for practitioners, not legal advice. Check the source text for your own obligations.