GPT-5.2 Codex-Max Launches To A Few Users

“GPT-5.2-Codex-Max,” a new model for Codex being tested by OpenAI, is already available to subscribers. OpenAI reportedly started releasing GPT-5.2 “Codex-Max” to a limited number of subscribers in January 2026. This model is a more sophisticated version of the GPT-5.2-Codex family that was created especially for challenging software engineering jobs and agentic coding.

Though It is not yet a universal deployment for all users, however it is currently visible for some Codex platform premium customers. By asking the Codex interface directly which identifies itself as “GPT-5.2 Codex-Max” users have verified the model’s existence. 

The basic GPT-5.2-Codex is normally priced similarly to the base GPT-5 models, with notable discounts for cached input tokens, while particular “Max” tier pricing specifics are still being worked out. 

When some users ask Codex what model it is using in which they notice a new model called GPT-5.2-Codex-Max.

In December, OpenAI released Codex with GPT-5.2; however, premium users were not given access to a “max” variant at that time.

When OpenAI released GPT-5.2-Codex, it demonstrated that its agentic tool can manage significant changes like refactors and migrations without losing the plot, stay on course for lengthy tasks, and maintain a big repository’s context useable via compaction.

Additionally, Codex 5.2 appears to be more dependable when using tools, performs better on Windows processes, and can comprehend screenshots, UI problems, and schematics you give during a coding session thanks to its improved vision.

The pattern indicates that a 5.2 “Max” might be the next bump if you consider how GPT-5.1-Codex-Max sat above the standard versions and produced another jump.

In the upcoming days, OpenAI may release information about GPT-5.2-Codex-Max.