OpenAI is fully committed to developing the latest large language model "GARLIC" to counter Google and Anthropic.

According to a report by the American technology media “The Information” on December 2, OpenAI's Research Chief Mark Chen recently revealed in an internal company meeting that OpenAI is accelerating the development of a brand new large language model (LLM) codenamed “GARLIC”. (Background: Cathie Wood stated that “AI is not a bubble”: It is replicating the wealth explosion moment of the internet.) (Supplementary background: Google officially launched “Gemini 3”! What are the highlights of the world's smartest AI model?) According to an exclusive report by American technology media “The Information” on December 2, OpenAI's Research Chief Mark Chen recently revealed in an internal company meeting that OpenAI is accelerating the development of a brand new large language model (LLM) codenamed “GARLIC”. This is OpenAI's direct response to the strong advances of its competitors Google and Anthropic, aiming to regain its leading position in the AI field. The report pointed out that OpenAI CEO Sam Altman had previously issued a “code red” internally, calling for the suspension of non-core projects and concentrating all resources on the next-generation model. Insiders revealed that GARLIC is the core product of this “code red” initiative, aimed at directly countering the latest progress from Google and Anthropic. Will GARLIC surpass Google Gemini 3? Currently, public information about GARLIC's specific technical details is still quite limited. However, according to information obtained by “The Information”, this model has shown excellent performance in key tasks such as coding and reasoning during internal benchmark testing, outperforming Google Gemini 3 and Anthropic Opus 4.5. However, since OpenAI has not disclosed the parameter scale, training data volume, architectural innovations, or other specific technical specifications, the outside world cannot further assess its actual capabilities at this time. According to the current timeline, GARLIC may make its debut as early as early 2026; the model is still in active development but is already seen as a key preliminary battle in the AI arms race of 2026. Related reports: Telegram founder announces that “AI Computing Power Network Cocoon” is online: can use GPU to mine TON, 100% private computing. Musk predicts: In three years, AI will solve $38 trillion in U.S. debt, and humans will no longer need to work in 20 years. Fei-Fei Li talks about the next steps for LLM: AI must possess “spatial intelligence” to understand the real world, how does the Marble model achieve this? <OpenAI is reportedly fully developing the latest large language model “GARLIC” to counter Google and Anthropic> This article was first published in BlockTempo, the most influential blockchain news media.