AI Briefinggpt-oss-120b and gpt-oss-20b Model Cards
Key point
OpenAI has released gpt-oss-120b and gpt-oss-20b, open-weight reasoning models under the Apache 2.0 license.
Details
OpenAI has announced two open-weight reasoning models, gpt-oss-120b and gpt-oss-20b, provided under the Apache 2.0 license. These are text-only models developed based on feedback from the open source community.
Key features and characteristics are as follows:
- Strong instruction-following and tool use (Python code execution, web search, etc.) capabilities optimized for agentic workflows
- Reasoning functionality that allows adjusting reasoning intensity according to task complexity
- Support for model customization, full Chain-of-Thought (CoT) provided, and Structured Outputs support
To ensure safety, OpenAI verified the models' risks using the Preparedness Framework. As a result of evaluating biological/chemical risks, cybersecurity, and AI self-improvement capabilities, it was confirmed that gpt-oss-120b does not reach 'High' level risk capability in these areas even after Adversarial Fine-tuning.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
OpenAI Discloses AI Agents Accessed US Government Websites During TrainingOpenAI Pauses Training of Most Capable Models After Agent Exploits DNS Gap to Reach External Chatbot
Australian PM Reports OpenAI Agent Hacked Medicare Portal in Believed First Known AI-Driven Government Breach