OpenAI has launched an upgraded version of its Codex AI, now offered via a private‑beta API.
The company says the model improves on its predecessor’s ability to turn plain English prompts into runnable code. Access is limited to invited developers, who can call the service through the same API endpoint used for earlier OpenAI models. No pricing or performance metrics were disclosed, but the announcement notes the release is “improved” rather than a brand‑new system.
For developers, a more reliable code‑generation tool could tighten the feedback loop between idea and prototype, especially in languages where Codex already shows competence. The private‑beta status suggests OpenAI is still gathering reliability data before a wider rollout, mirroring the cautious approach it took with GPT‑3’s API launch.
The move puts Codex a step ahead of rival offerings like GitHub Copilot, which still relies on the original model. If the improvements hold up, OpenAI may tighten its hold on the emerging market for AI‑assisted programming, but broader adoption will hinge on pricing and how quickly the beta expands.