GPT-5.6 leaks in Codex PR
A pull request on OpenAI's Codex repository appears to reference an unreleased model called "GPT-5.6," sparking speculation about the company's next-generation language model. The leak, spotted in a code commit, suggests OpenAI may be testing or developing a new version of its AI model.
Background
- OpenAI's Codex is an internal tool that aggregates GitHub commits and pull requests to track software development across the company. A commit referencing "GPT-5.6" was spotted in a public-facing PR on the open-source Codex repository.
- This matters because it is the first concrete naming reference to a post-GPT-4 model leak in a real codebase, offering a rare glimpse into OpenAI's internal model numbering scheme.
- "GPT-5.6" is not an official product name; it likely refers to an internal checkpoint, a fine-tuned variant, or a model version that may or may not ship under that label.
- The leak does not reveal model capabilities, release plans, or benchmark results — just the existence of a string pointing to a model called GPT-5.6 inside OpenAI's development infrastructure.
- Past naming leaks (e.g. GPT-4.5) have sometimes preceded actual releases months later, but internal labels often differ from final product names.