Search
OpenAI Unveils GPT-5.2-Codex, a Model That Shines in Software Engineering and Cybersecurity
2025-12-19 /
Read about 0 minute
Author：小编
On Thursday, OpenAI rolled out the GPT-5.1-Codex-Max intelligent agent coding model, specifically for its paid ChatGPT subscribers. In the SWE-Bench Pro test, this model attained an impressive accuracy rate of 56.4%, outperforming GPT-5.2, which scored 55.6%. OpenAI emphasized that the newly released model has witnessed a substantial enhancement in its cybersecurity proficiency. While it hasn't yet attained the 'high' - level benchmark, the company is gearing up for upcoming models to break through this barrier.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
