Sign In
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
New Delhi: OpenAI is moving quickly with a new large language model called Garlic as competition in the AI space grows hotter. The Information first reported that Garlic has been performing strongly in internal evaluations. It is being positioned as a direct answer to Google’s new Gemini 3 model, which recently gained strong public attention for its improvements in coding and reasoning.
Inside OpenAI, the pressure is clearly rising. Reports say CEO Sam Altman has already declared a “code red” to improve ChatGPT quality. This new push looks like a serious attempt to regain momentum immediately.
The report said OpenAI’s Chief Research Officer Mark Chen told colleagues that Garlic is already doing well when compared to Gemini 3 and Anthropic’s Opus 4.5 on tasks such as reasoning and coding. A source told The Information that OpenAI plans to release a version of Garlic as soon as possible. It could come to users as GPT 5.2 or GPT 5.5 by early next year.
The interesting part here is timing. Gemini 3 was only launched recently. OpenAI does not want the gap to widen. Altman told colleagues that OpenAI has a new reasoning model ahead of Gemini 3 in internal evaluations.
Garlic is not the only new model under work. There is another one earlier called Shallotpeat. The Information mentioned that Garlic includes bug fixes that OpenAI applied to Shallotpeat during pretraining. Pretraining is the stage where a model learns from large amounts of online and licensed data. Google recently highlighted new advancements in that particular step while building Gemini 3.
Mark Chen told staff that improvements in pretraining helped OpenAI overcome big issues they were having before. He said the company improved upon its previous best and much larger pretrained model GPT 4.5. That model launched in February and almost disappeared from the spotlight soon after.
Chen reportedly noted that these improvements allow OpenAI to infuse a smaller model with the same level of knowledge that earlier required a much larger model. That is a huge factor in AI development. Bigger models usually mean higher cost and more time.
Chen also said OpenAI has already started working on a bigger and better model using what they learned from Garlic. So Garlic is like a stepping stone in the company’s race to stay ahead.
Even with all this excitement, Garlic is not ready yet. The model still has to go through standard phases like:
All of that needs to be cleared before users can try it.
There is a moment happening in AI. Google confidently rolled out Gemini 3. Anthropic is moving ahead with Opus series. Users keep demanding smarter and more accurate chatbot responses. Startups want better reasoning for automation work. And developers online love comparing benchmarks just for fun.
OpenAI created the original buzz with GPT 3 and GPT 4. Now, Garlic is expected to be its next serious comeback. If OpenAI can deliver improvements fast, the fight between Gemini and Garlic could make every AI tool we use feel sharper.
