GPT-5-Codex: 7+ Hours Autonomous, Reviews Large Projects
GPT-5-Codex: OpenAI's autonomous coding agent for large-scale refactoring & review. Runs 7+ hours independently. Optimized for real-world engineering.
"AI Disruption" Publication 7700 Subscriptions 20% Discount Offer Link.
OpenAI has released GPT-5-Codex.
As the name suggests, this is a GPT-5 specifically optimized for intelligent agent programming tasks. OpenAI's blog states that GPT-5-Codex's training focuses on real-world software engineering tasks. It can respond quickly in short interactive conversations and independently complete lengthy, complex tasks. Its code review capabilities can identify critical vulnerabilities before code goes live.
GPT-5-Codex has been deployed across all Codex usage scenarios, including Codex CLI, IDE extensions, web interface, mobile devices, and code review in GitHub.
It is the default model for cloud tasks and code review, and developers can also choose to use it for local tasks through Codex CLI or IDE plugins. By the way: Codex is included in ChatGPT's Plus, Pro, Business, Edu, and Enterprise subscriptions.
Two and a half hours after release, OpenAI CEO Sam Altman excitedly announced that GPT-5-Codex traffic had already reached around 40% of Codex traffic, and he predicted the model would become the majority of Codex traffic within the day.