OpenAI has announced GPT-5.3-Codex. The new version aims to go beyond merely generating code, becoming a faster and more comprehensive assistant throughout the entire software development process.
OpenAI has announced its next-generation model focused on software development processes, GPT-5.3-Codex. Positioned by the company as a boundary-pushing coding model, the new version aims not only to generate code but also to play an active role throughout the entire software development lifecycle. Meanwhile, OpenAI's biggest rival, Anthropic, also announced that it is moving in the same direction with Opus 4.6.
GPT-5.3-Codex is accessible via command line, IDE plugins, a web interface, and a new macOS desktop application. API access is not yet open, but OpenAI states that it will be provided in the future.
According to OpenAI's internal test results, GPT-5.3-Codex outperforms both GPT-5.2-Codex and the general-purpose GPT-5.2 model. The new model achieved higher success rates in benchmark tests such as SWE-Bench Pro, Terminal-Bench 2.0, and similar evaluations. These results indicate that Codex operates more consistently and efficiently, especially in real-world software development scenarios.
Clarification on "Self-Developed" Claims
OpenAI has provided a clear clarification regarding recent headlines claiming "Codex developed itself." The company implicitly suggests that statements about GPT-5.3-Codex building itself are not true. However, it is not concealed that the model was actively used in its development process. According to OpenAI, GPT-5.3-Codex served as an important tool in its own development process but did not create itself in a fully autonomous manner. This approach is similar to AI-powered systems used by some large software firms today for deployment management, debugging, and test evaluation processes.
The most striking aspect of this update is OpenAI's positioning of Codex not merely as a code generation tool, but as an assistant encompassing the entire software development lifecycle. The company aims for GPT-5.3-Codex to be used in debugging, deployment, monitoring, writing product requirement documents, text editing, user research, testing, metrics, and many other areas. Furthermore, the model's ability to be guided during tasks and provide frequent status updates is also among its prominent features.
No ChatGPT Update Yet
Another notable detail is that ChatGPT is still on version 5.2. Codex's transition to version 5.3 creates expectations that OpenAI might plan a similar update for its general-purpose ChatGPT model. However, the company has not yet made an official announcement regarding this.
On the other hand, OpenAI has not announced any changes regarding limits or pricing with the new model. Nevertheless, a significant performance increase is being offered to Codex users. According to the company, improvements in infrastructure and inference systems allow the models to operate approximately 25 percent faster.
The company plans to transform Codex from merely a code-writing tool into one that can actively use a computer and complete real-world tasks end-to-end. This approach aligns with some existing solutions using methods like MCP, and also with systems similar to Anthropic's recently announced Claude Cowork.
0 Comments: