← Back to Examples
Large Language Model (Foundation Model)
GPAIA general-purpose AI model with 70 billion parameters, trained on large-scale internet data. Designed to be integrated into downstream applications by third-party providers.
Classification Walkthrough
- Step 1 — Prohibited practices (Article 5): The model itself does not perform any prohibited practice. ✗ Not prohibited.
- Step 2 — GPAI: The model is a general-purpose AI model as defined in Article 3(63) — it displays significant generality, is capable of competently performing a wide range of distinct tasks, and can be integrated into a variety of downstream systems. ✓ GPAI detected.
- Step 2a — Systemic risk: Training compute of 10^23 FLOPs is below the 10^25 threshold, and the model has not been designated as posing systemic risk by the AI Office. ✗ Not systemic.
- Step 2b — Open source: The model is not released under an open-source licence. Standard GPAI obligations apply in full.
- Result: GPAI (Standard) — The model provider must comply with Articles 51-53: prepare technical documentation, provide information to downstream providers, establish a copyright compliance policy, and publish a training data summary. Enforcement date: August 2, 2025.