GPT-5.3-Codex

OpenAI’s premier agentic model for end-to-end software development and complex computer-use tasks.

OpenAI released GPT-5.3-Codex on February 5, 2026, as a category shift from simple code generation to autonomous software engineering. This model delivers 25% faster interactions and is the first to earn a High capability rating for cybersecurity under the Preparedness Framework. It demonstrates elite performance on benchmarks like SWE-Bench Pro and Terminal-Bench 2.0 (scoring 77.3% on command-line tasks). Engineers used early versions to debug the model’s own training infrastructure and manage its deployment. Access is currently live for paid users via the Codex app, CLI, and IDE extensions, with a programmatic API rollout following the initial safety pilot.

https://openai.com/index/introducing-gpt-5-3-codex/