OpenAI Codex vs GitHub Copilot, which is the best?
--------------------------------------------------
AI code completions are now pretty mature and thus very useful. With the help of them nowadays I don't look up to StackOverflow or similar services to solve certain problems. There are 2 AI completion services that I have access to, OpenAI Codex and GitHub Copilot. Both work pretty well and are quite fast. The main feature that I like about Copilot is that it is integrated very well into VSCode. On the other hand, OpenAI Codex doesn't have good integration with VSCode. There are some unofficial extensions but they are not able to give suggestions as fast as Copilot. But if you don't use VSCode, OpenAI would be better since you have access to the API. Another plus side of Codex is that you can train models with your code, because of that you'll have more accurate answers. The downside of Codex is will be paid when it's out of private beta. I think it's the best to use copilot if you want something easy and use VSCode. If you want to use API's, train your own models and willing to pay later then use Codex.