Follow ZDNET: Add us as a favorite source On Google.
ZDNET Highlights
- OpenAI felt the pressure of recent OpenAI and Anthropic releases.
- CEO Sam Altman has reportedly initiated a “Code Red”.
- As a result, OpenAI is working on a new “garlic” model.
After Google released Gemini 3, which quickly rose to the top LMArena tops AI leaderboardOpenAI CEO Sam Altman informed employees that he was declaring a “Code Red”. According to a report from, the aim was to further improve ChatGPT to better compete InformationNow, a follow up report The publication reveals that the company is developing a new model in response, codenamed Garlic.
(Disclosure: ZDNET’s parent company Ziff Davis filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in the training and operation of its AI systems.)
Also: Is DeepSeek’s new model the latest setback for proprietary AI?
According to the report, Mark Chen, chief research officer at OpenAI, informed colleagues that Garlic performed well in the company’s evaluation compared to Gemini 3 and Anthropic’s Opus 4.5 in tasks involving coding and logic. This is important because both Gemini 3 and Anthropic Opus 4.5, released last month, set new industry standards, with the former leading the way in logic and the latter leading in coding.
OpenAI did not immediately respond to a request for comment.
Chen also said that when developing Garlic, OpenAI addressed issues of pretraining, the initial stage of training in which the model begins to learn from a huge dataset. The company focused on broad connections before training the model for more specific tasks. According to Chen’s comments quoted in the report, these changes to pretraining enable OpenAI to put as much knowledge into a smaller model that was previously reserved for larger models.
Smaller models can be beneficial to developers because they are generally cheaper and easier to deploy – French AI lab Mistral emphasizes this with its latest release this week. For the company behind it, it is cheaper to build and deploy a smaller model. According to a previous article, garlic should not be confused with shallotpeat, a model Altman announced to its employees in October. reportInformation, which was also intended to fix bugs in the pretraining process.
Also: Amazon says new DevOps agents don’t need babysitting — you can try them here
As far as when to expect the model, Chen kept details vague and only said “as soon as possible” in the report. However, given the context and the urgent need for OpenAI to stay ahead, it would be safe to assume that the model could be released as early as next year. Developments made while creating the garlic have allowed the company to develop its next bigger and better model, Chen said.
A fight for users
This fierce race between Google and OpenAI can partly be attributed to both competing for the same territory: consumers.
As CEO of Anthropic, Dario Amodei noted in conversation with Andrew Ross Sorkin the new York Times‘ At the DealBook Summit Wednesday, Anthropic isn’t in the same race or experiencing “Code Red” jitters as its competitors, because it’s focused on serving enterprises rather than consumers. The company has just announced that its cloud code agentic coding tool reached $1 billion in run-rate revenue just six months after being available to the public.
