OpenAI has introduced a major upgrade to its agentic coding tool, Codex, with the launch of a new model known as GPT-5.3 Codex.

The announcement came just minutes after rival AI company Anthropic released its own agentic coding model, highlighting the intensifying competition in the artificial intelligence development space.

According to OpenAI, GPT-5.3 Codex significantly expands Codex’s capabilities beyond basic code writing and review.

The company says the new model can now perform nearly all computer-based tasks carried out by developers and professionals, making it easier for a wider range of users to build software and automate complex workflows.

OpenAI claims the model has demonstrated strong performance across internal benchmarks, including the ability to create sophisticated games and applications from scratch over several days.

The company also noted that GPT-5.3 Codex operates about 25 percent faster than its predecessor, GPT-5.2, improving efficiency for large-scale projects.

Notably, OpenAI described GPT-5.3 Codex as its first model to meaningfully assist in its own development. Early versions of the system were used by engineers to debug the model and assess its performance, marking a milestone in self-assisted AI development.

The release was closely timed with Anthropic’s launch after both companies initially planned simultaneous announcements.

Anthropic ultimately moved its release 15 minutes earlier, briefly gaining the spotlight in what has become a highly competitive race to dominate AI-powered software development tools.

Source: Techcrunch