AI News (2026/2/6): OpenAI Releases New Programming Model: GPT-5.3-Codex
Executive Summary:
On February 6, OpenAI released the new programming model GPT-5.3-Codex, which achieved State-of-the-Art (SOTA) results in the SWE-Bench Pro and Terminal-Bench 2.0 tests, with a programming score 11.9% higher than Claude Opus 4.6. GPT-5.3-Codex has the ability to debug, deploy, and operate office software, with a 25% speed improvement and the capability to participate in its own development optimization.
OpenAI Releases New Programming Model: GPT-5.3-Codex Enhances Programming Efficiency
News Details
On February 6, OpenAI released its latest programming model, GPT-5.3-Codex. The model performed exceptionally well in multiple programming benchmark tests, particularly achieving State-of-the-Art (SOTA) results in the SWE-Bench Pro and Terminal-Bench 2.0 tests. Compared to Claude Opus 4.6, GPT-5.3-Codex improved its programming score by 11.9%. Additionally, the model excels not only in code generation but also in debugging, deployment, and operating common office software such as Microsoft Office and Google Workspace.
Key Points
Performance Improvement: GPT-5.3-Codex achieved 92.7% and 89.4% accuracy in the SWE-Bench Pro and Terminal-Bench 2.0 tests, respectively, significantly outperforming the previous generation of models.
Multifunctionality: In addition to code generation, GPT-5.3-Codex can also perform debugging, deployment, and operate common office software like Microsoft Office and Google Workspace.
Self-Optimization Capability: GPT-5.3-Codex can participate in its own development optimization process, continuously improving its performance through self-feedback mechanisms. This makes the model more flexible and efficient in practical applications.
AI-ALL In-Depth Commentary
The release of GPT-5.3-Codex marks another significant breakthrough for OpenAI in the field of programming models. Its outstanding performance in multiple benchmark tests not only validates the technical strength of the model but also provides developers with more powerful tool support. The enhanced multifunctionality means that the model can better adapt to complex development environments, boosting developer productivity. The introduction of self-optimization capabilities offers a new path for continuous improvement, allowing the model to evolve in practical applications. These advancements challenge existing programming assistance tools and open up new possibilities for AI in software development.
