The Chatgpt text does a very good job of generating and breaking the code from the prompt. Now, Openai has added a new coding agent to the chat, and it is not just a programmer who should be excited
Chatgpt’s Codex takes AI programming to the next level
Openai Codex is starting a research preview, which is a “cloud-based software engineering agent”. The feature is powered by Codex -1, a version of the OpenAII O3 model that is adapted for coding and software engineering works. Codex -1 is also trained to closely align its output with “human coding preferences and standards”.
If you are a chatgpt pro, enterprise, or team user, you can get the facility in the chat sidebar, in which plus and EDU users are getting it soon. Once opened, you can either assign a coding task by typing it a prompt and entering it. Code Ask or ask questions about button or your codebase Ask button. You will get information about the work list of Codex and progress below the Prompt Bar.
This new agent can do many tasks on the existing codebase, such as adding new features, fixing the bug, and you may have to answer any questions. Each task runs in a separate separate environment, which is preloaded with your codebase or repository. Codex can also read and edit files.
Declaration of openi Claims that the agent will take anywhere between one to 30 minutes to complete an assigned task, depending on the complexity of the work. You can monitor its progress in real time or even run many tasks together, all using their browser and computer as usual.
While Chatgpt can help you generate codes and even provide complete projects that you can download and test, it does not work well with software repository and codebase. Codex capabilities within typical software engineering infrastructure means that it is much more useful than vanilla chats, for both companies and individuals who maintain many projects in repository.
Codex produces cleaner codes compared to Codex Chatgpt, which is ready for human reviews and integration in workflows or codebase. It also conducts the test until it passes all the given test cases and conditions. Once a task is completed, the Codex will change its environment and “provide verification of its functions through the quotes of the terminal log and test output.”
Why is Codex a big deal?
Codex is a big thing for professionals in any industry. You can write Excel Macrose, report, report, edit batch, and do everything that will require essential expertise in some programming or scripting language.
Certainly, Chatgpt can generate codes and scripts for you, but in my experience, it is not reliable. You should have a relatively good knowledge of the programming language with which you are working and the code is a common idea of ​​debugging. Codex, however, automatically examines its code and runs the test to ensure that it works in your intended manner.
This can improve with the new GPT-4.1 model of Chatgpt, but this is not a perfect solution. To know which Chatgpt model to use, it can largely affect the output, so a model for coding will perform better than a more general-purpose model.
Of course, if you are a programmer, Codex is largely helpful as it can integrate with your github repository and take care of repeated functions and testing cases. This lets you grow and ship your app faster without getting caught in maintenance, tests and other functions, usually part of the software development process.
