OpenAI has released version 0.23 of Codex CLI, which now includes a 50% increase in usage limits for ChatGPT Plus subscribers. The updated Codex CLI integrates GPT-5, offering enhanced coding capabilities and improved reasoning features. Users can access GPT-5 with high reasoning effort through command-line shortcuts, and a new "/model" command has been added to facilitate model selection. While some users praise GPT-5's performance in coding and clinical reasoning—highlighted by a 17-point improvement over GPT-4o and a 27% advantage over human medical professionals—others have criticized the Codex CLI interface itself as less robust for coding tasks. GPT-5-mini, a variant optimized for agentic tasks such as writing, running, and error-correcting code in real time, demonstrates advanced resourcefulness beyond previous models. The integration of GPT-5 into Codex CLI is included in the ChatGPT subscription plan, making advanced AI coding tools more accessible to developers.
GPT-5 everywhere 🔥 https://t.co/oMJqD7IROG
gpt-5 high reasoning via codex cli is passing the sniff test
GPT-5 outperformed 4o by 17 points and human medical professionals by 27% regarding clinical reasoning. The trend is clear and continues to grow. Doctors who do not AUGMENT themselves with AI could be considered malpractice.