OpenAI Unveils GPT-5.3 Codex Minutes After Anthropic Release
▼ Summary
– OpenAI launched GPT-5.3 Codex, a new model designed to significantly enhance its Codex agentic coding tool for developers.
– The company claims the upgraded model can now perform nearly any task a developer does on a computer, enabling the creation of complex apps and games from scratch.
– OpenAI states GPT-5.3 Codex is 25% faster than its predecessor, GPT-5.2.
– The model is notable for being the first where early versions were used by staff to debug and evaluate the model itself.
– This release followed a competitive, nearly simultaneous launch by rival Anthropic, which moved its announcement up by 15 minutes.
The competitive landscape for AI-powered development tools just intensified with a significant new release. OpenAI has launched GPT-5.3 Codex, a model designed to supercharge its existing Codex agentic coding tool. This announcement arrived just minutes after its rival, Anthropic, unveiled its own competing agentic coding model, highlighting the fierce pace of innovation in this sector. OpenAI positions this upgrade as a transformative step, moving Codex beyond basic code generation and review into a more comprehensive professional tool.
According to the company, GPT-5.3 Codex enables the software to perform “nearly anything developers and professionals do on a computer.” This expanded capability aims to democratize software creation by changing who can build applications and how that work is accomplished. In performance testing, OpenAI asserts the model can construct highly functional, complex games and applications entirely from scratch within a matter of days, a claim that underscores its potential to accelerate development cycles dramatically.
Beyond new functionalities, the model boasts improved efficiency. OpenAI states GPT-5.3 Codex operates 25 percent faster than its predecessor, GPT-5.2. Perhaps more intriguing is the development process itself. The company notes this is its first model that played a key role in its own creation. Engineers utilized earlier iterations of the program to assist in debugging the system and evaluating its performance, a recursive approach to AI development.
The timing of this launch adds a layer of competitive drama. Both OpenAI and Anthropic had initially scheduled their announcements for the same precise moment. However, Anthropic adjusted its plan at the last minute, moving its release forward by fifteen minutes. This slight shift allowed Anthropic to technically announce first, though OpenAI followed almost immediately with its own substantial update. This back-to-back release of advanced coding agents signals a new phase in the race to provide the most powerful and practical AI tools for software engineering.
(Source: TechCrunch)
