OpenAI has released two open-weight reasoning models, gpt-oss-120b and gpt-oss-20b, under the Apache 2.0 license. These text-only models are designed for agentic workflows, supporting instruction following, tool use, and adjustable reasoning effort. OpenAI emphasizes that while the models adhere to safety policies by default, their open nature means users must implement additional safeguards, as they cannot revoke access or implement system-level protections like API-served models. Evaluations indicated that even with adversarial fine-tuning, gpt-oss-120b did not reach high capability thresholds in biological, chemical, or cyber domains, and its release is not expected to significantly advance frontier capabilities in open foundation models. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
RANK_REASON OpenAI released open-weight models with a model card, which falls under research releases rather than frontier releases.
