OpenAI is accelerating updates to ChatGPT in response to recent advancements from competitors. Reports from early last week indicate that CEO Sam Altman sent a “code-red” memo to employees, directing them to focus on ChatGPT development ahead of other initiatives. This internal push aims to strengthen the chatbot’s position against Google’s Gemini 3, which launched last month.
The memo, as detailed in multiple outlets, highlights the impact of new models from rivals on OpenAI’s leadership. A new reasoning model for ChatGPT could launch as soon as this week, according to these reports.
On Friday, The Verge reported that the anticipated upgrade is GPT-5.2, with Tuesday, December 9, set as a potential release date. The publication stated that GPT-5.2 is prepared for deployment this week as an initial counter to Gemini 3. However, OpenAI has not confirmed the timeline, and the launch could be delayed.
In a separate development, The Information revealed last week that OpenAI is working on a project codenamed “Garlic.” This update introduces a new architecture for ChatGPT, designed to rival Gemini 3 and Anthropic’s Claude Opus 4.5. Garlic might debut as GPT-5.2 or GPT-5.5, with an official rollout targeted for early 2026.
Google’s Gemini 3, released a few weeks ago, achieved significant performance improvements and led several AI benchmarks. Anthropic followed with Claude Opus 4.5, which also excelled in those tests. OpenAI introduced its GPT-5 models in August and upgraded ChatGPT to GPT-5.1 shortly before the Gemini 3 and Opus 4.5 launches.
Altman’s memo underscores how these competitor models influenced OpenAI’s strategy. The Verge noted that Garlic differs from GPT-5.2. Originally, OpenAI planned GPT-5.2 for later in December, but competitive pressures prompted an earlier timeline. If GPT-5.2 serves as the immediate reply to Gemini 3, it may not incorporate the full Garlic enhancements.
Details on GPT-5.2 emphasize practical upgrades over novel capabilities. The Verge described improvements in speed, reliability, and customizability to help OpenAI close the gap with Google. The Wall Street Journal, covering the code-red memo before The Verge’s date announcement, aligned on these enhancements. It reported that a new ChatGPT reasoning model is due this week, positioned to surpass Gemini 3 across various tasks. The Journal did not reference the GPT-5.2 designation.
Garlic brings a specific technical advancement: enhanced pretraining efficiency. This allows a smaller model to retain the knowledge capacity of a larger one, potentially increasing speed and lowering computing expenses. Benchmark evaluations show Garlic performing well, particularly in programming tasks and other standard AI metrics.
