OpenAI retires GPT-4o, moves fully to GPT-5.2
OpenAI has officially retired the GPT-4o model from its ChatGPT product, completing the full transition to GPT-5.2. The company also removed other older models to consolidate support and maintain a rapid iteration cadence. Some users have expressed dissatisfaction, mourning the loss of specific model behaviors they favored.
- The release of GPT-5.2 on December 11, 2025, was reportedly accelerated to compete with Google's Gemini 3 Pro and Anthropic's Claude 4.5, which had launched in the preceding weeks. - This was OpenAI's second attempt to retire GPT-4o; an initial effort in August 2025 was reversed after a user outcry that the new GPT-5 models lacked the "warmth and encouraging tone" of their predecessor. - For API users, GPT-5.2 is cheaper for input processing ($1.75 per million tokens vs. GPT-4o's $2.50) but more expensive for output generation ($14.00 per million tokens vs. GPT-4o's $10.00). - While removed from the consumer-facing ChatGPT product on February 13, 2026, the GPT-4o model remains available for developers via the API, with OpenAI promising advance notice before any future API retirement. - The new flagship model, GPT-5.2, features a 400K input token context window
