General‑purpose AI model

AI models designed and deployed to perform a broad range of tasks rather than a narrowly specified single purpose; the Act introduces obligations for providers of such models (technical documentation, transparency to downstream users and systemic‑risk mitigation) particularly when they are designated as posing systemic risks.

A model designed to perform a wide range of tasks across different contexts rather than being tailored to a specific legally defined purpose; providers of general‑purpose models must supply enhanced documentation and transparency about training sources and may face heightened oversight or designation by the AI Office if assessed as posing systemic risk.

A model intended to perform different tasks across multiple domains and use cases rather than a single narrowly defined purpose; such models are subject to specific transparency, robustness and additional obligations (and may be designated as posing systemic risk) under the Regulation when placed on the Union market or when their outputs are used in the Union.

A model that has been trained using compute exceeding 10^23 floating point operations and is capable of generating language (text or audio), text‑to‑image, or text‑to‑video outputs. The Guidelines use this threshold plus generative capability to determine which models fall within GPAI obligations under the AI Act.

A model intended for a broad range of tasks and adaptable across multiple use cases, which may be deployed or fine‑tuned for diverse functions; providers of GPAI face targeted obligations such as documentation, governance, transparency measures and potential EU‑level evaluations.

A model capable of performing a wide range of tasks across different contexts rather than being limited to a specific narrowly defined purpose; under the Regulation, providers of GPAI must meet specific transparency and record‑keeping obligations, cooperate with the EU AI Office, and may be subject to direct Union supervision if the model poses systemic risks.

A model, as adopted from the EU AI Act, that is not limited to a specific narrowly defined purpose but can be adapted to perform diverse tasks across different domains; under the Act GPAIs are subject to extended governance and transparency obligations, including additional documentation and reporting duties toward national and EU authorities.

AI models designed or marketed to perform a wide range of tasks and to be adaptable to many different purposes (as distinct from purpose‑specific systems); the Act imposes bespoke governance, information, reporting obligations and potential EU‑level supervision for providers of GPAI.