GPT-OSS 120B

Model ID: vllm/release/gpt-oss-120b OpenAI LTS

EU AI Act

Attribute	Value
Classification	GPAI (General-Purpose AI Model)
Risk class	Limited Risk (Article 52)
Transparency obligation	Yes
noris role	Deployer

As a GPAI model with systemic-risk potential, this model falls within the scope of EU AI Act transparency obligations. noris acts as the deployer and ensures the required information is made available pursuant to Article 53.

Technical Specifications

Property	Value
Model name	GPT-OSS 120B
Provider	OpenAI
Model ID	vllm/release/gpt-oss-120b
Parameters	117B (5.1B active, MoE)
Architecture	MoE Transformer
Context length	128K tokens
Modalities	Text only
Quantization	MXFP4
License	Apache 2.0
Release	August 2025

Recommended Use Cases

• Complex reasoning
• Code generation
• Agentic workflows (function calling, web browsing, Python execution)
• Academic research
• Enterprise document processing

Known Limitations

Caution

• Knowledge cutoff late 2024: the model has no knowledge of events after its training period.
• Text only (no vision/audio): multimodal input is not supported.
• Can show sycophancy/hedging: the model occasionally tends toward overly agreeable or evasive answers.

Risk Assessment (EU AI Act)

Risk Factor	Category	Level
Open source	Misuse potential	Medium
Model size	Deepfake potential	Medium
Training data	Copyright	Medium
Environmental impact	Model size	Medium

Training Data & Sustainability

Training data: web crawl, licensed content, synthetic data (through late 2024).

Sustainability:

Metric	Value
CO₂ (estimated)	~50 to 200 tCO₂e
Energy / 1K tokens	~0.05 to 0.2 kWh
Optimization	MXFP4 quantization for efficient inference

Original Sources

• HuggingFace: https://huggingface.co/openai/gpt-oss-120b
• License: Apache 2.0
• Model card: https://arxiv.org/abs/2508.10925

Model card created on July 2, 2026. This model card serves EU AI Act transparency purposes (Regulation (EU) 2024/1689). The information is based on publicly available data from the model provider.