gpt oss safeguard 120b

Name: gpt oss safeguard 120b
Author: OpenAI

by OpenAI

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released September 2025131K context≈ 98,304 words120b params

A specialist rather than a generalist, this model focuses narrowly on content safety classification and policy enforcement. It excels at flagging harmful, unsafe, or policy-violating content with the precision you'd expect from a dedicated auditor, but won't help you draft an email or debug code — that's simply not its job. Think of it as a focused colleague who only reviews content for safety concerns, doing that one thing with the weight of a 120B parameter model behind it.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Multimodal

Moderate

Creative Writing

gpt oss safeguard 120b

by OpenAI

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released September 2025131K context≈ 98,304 words120b params

A specialist rather than a generalist, this model focuses narrowly on content safety classification and policy enforcement. It excels at flagging harmful, unsafe, or policy-violating content with the precision you'd expect from a dedicated auditor, but won't help you draft an email or debug code — that's simply not its job. Think of it as a focused colleague who only reviews content for safety concerns, doing that one thing with the weight of a 120B parameter model behind it.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Multimodal

Moderate

Creative Writing

Glossary

ClassificationA machine learning task where the model assigns input data to predefined categories or labels.Content Safety ClassificationThe task of automatically detecting and categorizing text that violates policies or could cause harm, such as hate speech, violence, or misinformation.Parameter ModelA neural network described by the number of learnable weights it contains; more parameters generally mean greater capacity to learn complex patterns, but also require more computational resources.PolicyA learned model that maps observations to robot actions, trained on demonstration data.Policy EnforcementThe process of automatically checking content against a set of rules or guidelines and blocking or flagging violations.PrecisionThe level of numerical detail a model uses to represent its internal values; higher precision means more accurate calculations but requires more memory.Safety ClassificationA machine learning task that assigns content to categories based on whether it poses safety risks or harms.

Capabilities

Capabilities

Use Case Fit

Glossary