OpenAI Codex Discontinues GPT-5.2 and GPT-5.3-Codex on June 2
OpenAI developer @thsottiaux announced that OpenAI will discontinue the GPT-5.2 and GPT-5.3-Codex models in Codex on June 2.
For free plan users, GPT-5.5 will become the default cutting-edge model for future builds and usage.
The discontinued models will remain available on the OpenAI API.
Source: Public Information
ABAB AI Insight
OpenAI is rapidly iterating the GPT-5 series in the Codex coding tool, discontinuing older versions and setting GPT-5.5 as the default model for free users, continuing its path of optimizing computational resource allocation and user experience through version replacement, while maintaining API compatibility for enterprise clients.
On the capital front, OpenAI is concentrating core computational resources on GPT-5.5, enhancing overall user stickiness and paid conversion by making the latest model the default for free users, while the API continues to support older models to ensure a smooth transition for large clients.
Similar to previous model switching strategies in Codex, OpenAI is currently in a phase of efficiency optimization, moving from multiple versions coexisting to a unified leading-edge model.
Essentially, this represents capital concentration: discontinuing old models and defaulting to the latest version, with the mechanism being that the new model offers better cost-performance for coding tasks, forcing computational resources to concentrate on the leading model, while balancing enterprise revenue and cost control through API compatibility, driving the developer ecosystem to quickly migrate to the latest model.
ABAB News · Cognitive Law
Free users are always pushed towards the latest default model, while paid users enjoy compatibility.
Computational power is always concentrated on the strongest model, with old models only remaining in the API for transition.
Leading labs continuously push resources towards performance frontiers through version iteration.