OllamaFrom the builder
OpenAI gpt-oss-safeguard
Ollama partnered with OpenAI and the nonprofit ROOST to offer the gpt-oss-safeguard reasoning models for safety classification, in 20B and 120B sizes under the Apache 2.0 license. The models apply policies that users write themselves, expose their full reasoning to developers, and offer low, medium and high reasoning effort. OpenAI tested them on internal evaluations and external sets, including the moderation dataset from its 2022 paper and ToxicChat.
Keep in mind Raw CoT is meant for developers and safety practitioners. It’s not intended for exposure to general users or use cases outside of safety contexts.
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.