OpenAI Unveils GPT-5.3-Codex: The 'Most Capable Agentic Coding Model'
On February 5, OpenAI, led by Sam Altman, introduced a groundbreaking new Codex model named GPT-5.3-Codex. The company boldly claims this is the "most capable agentic coding model to date" and marks a significant milestone as the first model to "meaningfully contribute to its own creation." According to OpenAI's official statement, early versions of GPT-5.3-Codex have already been utilized internally for critical tasks such as debugging training processes, managing deployment protocols, and analyzing test results and evaluations.
What Can the Latest Codex Model Do?
GPT-5.3-Codex is designed to handle long, complex tasks that encompass research, tool use, and multi-step execution. A key feature is its ability to allow coders to interact with and guide the model in real-time, much like collaborating with a human colleague, without losing context. This interactive capability enhances productivity and streamlines development workflows.
Features of GPT-5.3-Codex
OpenAI detailed that the GPT-5.3-Codex model integrates the advanced coding prowess of the previous GPT-5.2-Codex with the strong reasoning and professional knowledge of GPT-5.2 into a single, unified system. Additionally, the company highlighted that the new model operates at least 25% faster than its predecessors, boosting efficiency for users.
What's New with GPT-5.3-Codex
With GPT-5.3-Codex, the model has evolved from a tool that primarily wrote and reviewed code into a versatile agent capable of performing nearly any task that developers and other professionals execute on a computer. In simpler terms, while earlier versions focused on coding, the upgraded model can now:
- Research information comprehensively
- Edit files and manage workflows
- Run various tools and applications
- Handle tasks across the entire software lifecycle
This includes activities such as debugging, deploying, monitoring, writing Product Requirements Documents (PRDs), editing copy, conducting user research, running tests, analyzing metrics, and even processing data in spreadsheets.
Availability of GPT-5.3-Codex
The latest GPT-5.3-Codex is available to all users with paid ChatGPT plans and can be accessed anywhere. This broad availability covers:
- The main application interface
- Command Line Interface (CLI)
- Integrated Development Environment (IDE) extensions
- Web-based platforms
Rival AI Model Launched by Anthropic
The launch of GPT-5.3-Codex coincided with Anthropic's release of its new AI model, Claude Opus 4.6, intensifying the rivalry between the two leading AI firms. Anthropic described Opus 4.6 as an improved version built on its predecessor, featuring stronger coding abilities. Key enhancements include careful planning capabilities, reliable handling of long-running agentic tasks, better performance in large codebases, and improved code review functions.
OpenAI vs Anthropic: A Competitive Landscape
Following the simultaneous launches, OpenAI's Sam Altman and Anthropic's leading researcher Sholto Douglas appeared on a podcast, as reported by Business Insider. Altman commented on the evolving AI landscape, stating, "I think we will be heading towards a workflow where a lot of people just feel like they're managing a team of agents. And as the agents get better, they'll keep operating at a higher and higher level of abstraction."
In response, Douglas noted key differences between the companies' approaches, saying, "The OpenAI models were a bit better at trying really, really, really hard on tough problems, but the Anthropic models were much faster and so forth." He added that while OpenAI has focused on improving speed, Anthropic has prioritized making AI models "much, much better at really, really tough problems."
This competitive dynamic underscores the rapid advancements in AI technology, with both firms pushing boundaries in coding and agentic capabilities. The developments signal a shift towards more autonomous and collaborative AI tools in software development and beyond.
