OpenAI Developing New LLM ‘Garlic’ as ChatGPT Enters Code Red Mode
OpenAI appears to be entering one of the most crucial periods in its history. While CEO Sam Altman is pushing the company into full “code red” to strengthen ChatGPT, new reports claim the AI pioneer is also developing another powerful large language model. Codenamed “Garlic,” this new model reportedly delivers strong performance in advanced reasoning and coding tasks, and has outperformed competitors like Google’s Gemini 3 and Anthropic’s Cloud Opus 4.5 in internal tests. This development marks a major strategic move by OpenAI at a time when the AI race has become more competitive than ever.
According to The Information’s reporting, Garlic isn’t just another upgrade to the GPT family. Rather, it reflects a shift in the company’s long-term vision—one that is strongly oriented toward specific, high-value domains like healthcare, biomedicine, and scientific research. While previous OpenAI models, including GPT-4 and GPT-5, focused on broad general intelligence, Garlic is being developed as a specialized model designed to excel at deep reasoning, complex logic, and domain-specific workflows.
The timing of this development is noteworthy. Both Google and Anthropic have made significant strides in the past few months. Google launched Gemini 3 and rapidly integrated it into its ecosystem, touting it as its fastest deployment to date. Anthropic introduced Cloud Opus 4.5 shortly after, marketing it as the world’s most capable coding model. In this regard, OpenAI’s work on Garlic is being seen as a significant response to significant market pressure.
Garlic Internal Performance: Outpacing Gemini 3 and Claude Opus 4.5
OpenAI’s Chief Research Officer, Mark Chen, reportedly told his colleagues that Garlic is performing very well in internal evaluations. The model appears to outperform Google’s Gemini 3 and Anthropic’s Cloud Opus 4.5 in tasks involving deep reasoning, coding precision, debugging, and multi-step logic generation. These are precisely the areas where OpenAI’s competitors have been gaining ground in recent months.
Chen’s involvement in Garlic’s development demonstrates the project’s significant importance within OpenAI. He has previously led major research initiatives such as the DALL·E, Codex, and o1 reasoning models, all of which have positioned OpenAI as a leader in multimodal and logic-focused AI. His leadership demonstrates that Garlic isn’t just another experimental project—it’s a strategic cornerstone in OpenAI’s roadmap.
If the leaks are accurate, Garlic is expected to arrive as GPT-5.2 or GPT-5.5, and is likely to be released in early 2026. However, given the growing competition, market analysts believe OpenAI could accelerate the timeline based on market pressure and technology readiness.
A Shift Toward Industry-Focused, Specialist AI
One of the most crucial aspects of Garlic is its underlying intent. Unlike past models that focused on general capabilities—conversation, content generation and broad knowledge—Garlic reportedly aims to specialise in high-value technical fields. Healthcare, biomedicine, software engineering and advanced scientific reasoning appear to be central focus areas.
This strategic pivot is important for the future of AI. As generalist models begin to converge in capability, the next wave of innovation will be driven by specialist systems that outperform humans in narrow domains. Companies in pharmaceuticals, biotech, scientific innovation and enterprise software are expected to be major beneficiaries of such models. OpenAI’s investment in Garlic shows that it understands the future of AI will not be dominated by general chatbots but by deeply integrated tools that accelerate specialised industries.
Competition Intensifies as OpenAI Declares ‘Code Red’
The buzz around Garlic coincides with Altman’s internal memo to employees, where he reportedly declared a company-wide “Code Red.” The memo urged teams to prioritise improvements to ChatGPT’s responsiveness, reliability and personalisation features. It also asked employees to delay several secondary initiatives, including upcoming advertising plans, so that resources could be fully re-allocated toward strengthening the flagship product.
The urgency is understandable. Google’s Gemini 3 rollout has been faster and more aggressive than any of its previous launches. The model performed strongly across image editing, text generation and multimodal reasoning benchmarks. Meanwhile, Anthropic’s Claude Opus 4.5 continues to dominate conversations among developers for its coding prowess.
For the first time since ChatGPT’s launch, OpenAI is facing real pressure on user growth. According to Palo Alto-based entrepreneur Deedy Das, ChatGPT’s unique daily active users dropped by nearly 6 percent in the two weeks following Gemini 3’s release. This is a significant trend for a company whose valuation—rumoured to approach $500 billion—depends heavily on consumer momentum and expansion.
OpenAI’s consumer business remains the engine behind its revenue growth. Any visible decline in user engagement has serious long-term implications. Das pointed out that the year-end revenue target of $20 billion ARR depends on strong daily activity, making this dip the first major red flag in the company’s history.
Google’s Rapid Rise: Gemini App Hits 650 Million Monthly Users
Adding to OpenAI’s concerns is the rapid rise of Google’s Gemini ecosystem. A report by Fortune revealed that the Gemini app reached 650 million monthly active users in October, up from 450 million just three months earlier. This surge shows how aggressively Google is integrating AI across Android, Search, Workspace and YouTube.
The moment feels historically ironic. Three years ago, it was Google that declared ‘Code Red’ in response to ChatGPT’s explosive rise. Now, the positions seem to have reversed, with OpenAI reacting to Google’s accelerating dominance.
Google’s Nano Banana Pro model—which gained attention for its image generation quality—serves as another reminder that the AI race is shifting quickly. OpenAI cannot afford a slow response, and Garlic may very well be their opportunity to regain an edge.
OpenAI Prepares a New Model Ahead of Gemini 3
Altman’s memo reportedly hinted that OpenAI plans to release a new model as early as next week, which the company believes is ahead of Gemini 3 in several performance categories. While details remain under wraps, the memo suggested improvements in model behaviour, image generation and reliability, areas where users have demanded better performance for months.
Given the timing, this release may serve as an interim upgrade while the larger Garlic model continues its development pipeline.
Internal Challenges and Research Leadership
Despite losing several senior researchers to rival companies this year, OpenAI still retains one of the largest and most experienced research teams in the world. Many of its top scientists remain deeply involved in frontier-model development, and Chen’s leadership is expected to play a pivotal role in closing the competitive gap.
The company’s challenge now is not just innovation but speed. As the AI race grows fiercer, the ability to ship reliable, high-performing models quickly will determine which company leads in the long term. Garlic, ChatGPT upgrades and future GPT iterations are part of an expanding strategy aimed at reclaiming momentum that OpenAI once enjoyed effortlessly.
A Defining Moment for OpenAI’s Future in AI
The development of the Garlic LLM and the declaration of Code Red represent a defining moment for OpenAI. The company finds itself at the intersection of immense opportunity and rising competitive pressure. If Garlic truly surpasses top models from Google and Anthropic in reasoning and coding tasks, it could mark a new era for OpenAI—one where the company leads not only in general AI but also in specialised industrial intelligence.
For now, the world waits to see how OpenAI responds to the shifting landscape. With new models on the horizon, a renewed focus on ChatGPT and an ambitious specialist model in the works, the next twelve months will determine whether OpenAI stays ahead or gets overtaken in the most competitive AI race in history.
Click Here to subscribe to our newsletters and get the latest updates directly to your inbox