<!-- headlesslaw {"release":{"id":"a4f2a0495fe55b47323883f82977a9b34ab66171ec9623e39925d3a7045b6c81","published":"2026-11-01T16:41:30Z","pinned":false},"limits":{"coverage":{"history":"from:2026-07-27","held_since":"2026-07-27","gaps":[],"covered":["AI Act L1"],"not_covered":["national guidance","court decisions"]},"freshness":{"release_id":"a4f2a0495fe55b47323883f82977a9b34ab66171ec9623e39925d3a7045b6c81","release_published":"2026-11-01T16:41:30Z","checked_at":"2026-10-20","checked_against":"EUR-Lex"},"truncation":null,"failures":[],"incomplete":[]},"page":{"path":"/ai-act/articles/51","url":"https://www.headlesslaw.com/ai-act/articles/51","title":"Article 51 — Classification of general-purpose AI models as general-purpose AI models with systemic risk","family":"l1_generated","work_id":"aiact","status":"in_force","force_status":"in_force","date_state":"current","requested_date":null,"as_of":"2026-11-02","applies_from":null,"application":null,"version_label":"Consolidated version of 27 July 2026","check":{"against":"EUR-Lex","on":"2026-10-20"},"check_result":"matched","official_url":"https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02024R1689-20260727","valid_from":"2026-07-27","valid_until":null,"revision":2,"content_id":"61e504780ec80af5","change_date":null,"nearest_versions":["2026-07-27"],"first_text_date":"2024-06-13","correction_applies":true,"versions":["2026-07-27"],"version_url":"https://www.headlesslaw.com/ai-act/articles/51/versions/2026-07-27","revision_url":"https://www.headlesslaw.com/ai-act/articles/51/versions/2026-07-27/rev/61e504780ec80af5","versions_url":"https://www.headlesslaw.com/ai-act/articles/51/versions","relations_url":"https://www.headlesslaw.com/ai-act/articles/51/relations","source_line":null,"attribution":null}} -->
---
title: "Article 51 — Classification of general-purpose AI models as general-purpose AI models with systemic risk"
type: regulation
legal_status: in_force
source: "https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02024R1689-20260727"
source_retrieved: 2026-09-28
shortTitle: "Article 51 — Classification of general-purpose AI models as general-purpose AI models with systemic risk"
instrument_type: level1_regulation
authority: Commission
binding: true
binding_character: binding
mica_level2: false
legal_basis: "Regulation (EU) 2024/1689"
instrument: aiact
celex: "32024R1689"
regulation: AI Act
regulation-full: "Regulation (EU) 2024/1689"
article: 51
status: in_force
object_id: eu.aiact.legal_text.article.51
text_version: "Consolidated version of 27 July 2026"
source_url: "https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02024R1689-20260727"
check_status: matched
verified: true
last_checked: 2026-10-20
---

*In force · Consolidated version of 27 July 2026 · Checked against EUR-Lex on 20 Oct 2026 · [Official source](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02024R1689-20260727)*

1. <a id="p-1"></a>A [general-purpose AI model](https://www.headlesslaw.com/definitions/ai-act/general-purpose-ai-model) shall be classified as a general-purpose AI model with [systemic risk](https://www.headlesslaw.com/definitions/ai-act/systemic-risk) if it meets any of the following conditions:

    <a id="p-1-a"></a>**(a)** it has high impact capabilities evaluated on the basis of appropriate technical tools and methodologies, including indicators and benchmarks;

    <a id="p-1-b"></a>**(b)** based on a decision of the Commission, ex officio or following a qualified alert from the scientific panel, it has capabilities or an impact equivalent to those set out in point (a) having regard to the criteria set out in Annex XIII.
2. <a id="p-2"></a>A general-purpose AI model shall be presumed to have high impact capabilities pursuant to paragraph 1, point (a), when the cumulative amount of computation used for its training measured in floating point operations is greater than 10<sup>25</sup>.
3. <a id="p-3"></a>The Commission shall adopt delegated acts in accordance with [Article 97](97.md) to amend the thresholds listed in paragraphs 1 and 2 of this Article, as well as to supplement benchmarks and indicators in light of evolving technological developments, such as algorithmic improvements or increased hardware efficiency, when necessary, for these thresholds to reflect the state of the art.

***

[Chapter V — General-purpose AI models](../chapters/v.md) · [← Article 50](50.md) · [Article 52 →](52.md)
