OpenAI reduces Codex Model Context Size from 372k to 272k
OpenAI has decreased the context window size of its Codex model from 372,000 tokens to 272,000 tokens, as evidenced by a pull request on the official Codex GitHub repository. The pull request, titled "Reduce Codex Model Context Size from 372k to 272k," modifies a file within the repository to implement this change. The exact rationale for the reduction is not provided in the source text, but such adjustments typically impact the model's capacity to process large codebases or lengthy code generation tasks in a single pass. Codex is a specialized AI model designed for code generation and understanding, often used in development environments. This context size reduction may influence how developers interact with the model, particularly for tasks requiring extensive context, such as analyzing entire files or repositories. The change was published on July 19, 2026, and is publicly visible on GitHub. No further details about performance implications or future updates are mentioned in the source.
Smaller context may limit Codex's ability to handle large codebases in one request.