OpenAI is developing a new large language model codenamed “Garlic” that outperforms rivals in coding and reasoning tasks, as the company faces mounting pressure from competitors Google and Anthropic in the artificial intelligence race.
Garlic Takes Shape#
Chief Research Officer Mark Chen informed colleagues last week about the model, which is performing well in internal evaluations compared to Google’s Gemini 3 and Anthropic’s Claude Opus 4.5, particularly in coding and reasoning capabilities. OpenAI aims to release Garlic as soon as possible, potentially as GPT-5.2 or GPT-5.5 by early 2026.
Chen, who was elevated to Chief Research Officer in March 2025, has led development of major OpenAI projects including DALL-E, Codex, and the o1 reasoning models. His announcement signals OpenAI’s determination to maintain its position as competition intensifies from well-resourced rivals.
Altman Declares Code Red#
The development comes as CEO Sam Altman declared a “code red” on Monday, December 1, urging employees to prioritize improvements to ChatGPT while delaying other initiatives including advertising plans. In an internal memo, Altman said OpenAI would marshal more resources toward enhancing the chatbot’s responsiveness, reliability, and personalization features as competitive threats intensify.
Altman’s memo indicated OpenAI plans to release a new reasoning model next week that internal evaluations show is “ahead” of Gemini 3. The company will also focus resources on improving model behavior and image generation capabilities.
Shifting Competitive Landscape#
The urgency reflects a dramatic shift in the AI landscape. Google released Gemini 3 on November 18, deploying it across its ecosystem including Search in what the company called its fastest-ever rollout. The model quickly topped benchmark leaderboards for text generation, image editing, and multimodal reasoning. Anthropic followed with Claude Opus 4.5 on November 24, touting it as the best model for coding and autonomous agents.
Google’s Gemini app has grown to 650 million monthly active users as of October, up from 450 million in July, narrowing the gap with ChatGPT’s 800 million weekly users. The reversal is striking: three years ago, Google declared its own “code red” in response to ChatGPT’s launch, with CEO Sundar Pichai warning the chatbot could threaten Search.
OpenAI must also navigate the departure of dozens of top researchers to competitors including Meta and new startups, adding pressure to deliver results quickly.