OpenAI announced on the 6th that it has launched GPT-5.3 Codex, an enhanced agent-based coding model.
GPT-5.3 Codex combines the coding performance of GPT-5.2 Codex with the reasoning and specialized knowledge capabilities of GPT-5.2, while being 25% faster. Benchmark evaluations showed the model achieved industry-leading performance on SWE-Bench Pro and Terminal Bench. It also demonstrated strong results on OSWorld and GDPVal. These four benchmarks are metrics OpenAI uses to measure coding capabilities, agent capabilities, and real-world task performance.
GPT-5.3 Codex is also the first model to play a significant role in its own development. OpenAI's Codex team used early versions to debug its own training, manage deployment, and diagnose test results and evaluations. GPT-5.3 Codex is available to paid ChatGPT users in all environments where Codex is accessible. OpenAI also plans to provide secure API access for integration.
This update also aligns with recent partnership-based integrations. Through these integrations, developers can directly utilize OpenAI's Codex within Apple's Xcode and GitHub workflows to autonomously handle complex tasks.