OpenAI Project ‘Garlic’ Codename: Early Tests Show Strong Performance Against Rivals

OpenAI Garlic LLM project: OpenAI is said to be working on a new large language model codenamed Garlic, signalling the company’s push to stay ahead in a fast-moving AI race. The Information reports that Mark Chen, OpenAI’s chief research officer, privately shared details of Garlic with employees, and early internal testing suggests it’s delivering strong results in coding and reasoning. Performance comparisons reportedly show it stacking up well against Google’s Gemini 3 and Anthropic’s Opus 4.5 — two of the strongest models currently in public use.

OpenAI Project Garlic Codename: GPT 5.2/5.5 in Action

The model is expected to take shape as a mid-cycle upgrade rather than an entirely new generation. Garlic could land under the GPT-5.2 or GPT-5.5 branding, with suggestions pointing to an early 2026 launch window. If accurate, this timeline indicates OpenAI wants the model out quickly, rather than waiting for a full GPT-6 milestone.

ALSO READ: Government Orders To Pre-Install Sanchar Saathi App On Smartphones: Are There Any Privacy Concerns?

Internal ‘Code Red’ Push at OpenAI

Rumours of Garlic arrive amid what Sam Altman has reportedly called a “code red” phase. According to The Wall Street Journal and The Information, Altman told teams that competition is heating up fast, and OpenAI must deliver sharper improvements to ChatGPT. He has urged the company to prioritise speed, reliability, accuracy, and more personalised user experiences.

Multiple internal projects have been paused so resources can be redirected. These include experimental ad tools, health-orientated AI, shopping agents and Pulse — a personal assistant concept. Daily check-ins are also expected for teams handling the upgrades, and employees from other areas are being encouraged to temporarily shift over to boost progress.

ALSO READ: Google Android 16 Update: Brings AI Notification, And Summaries, Checkout Eligible Devices

Why Garlic Matters

Google, Meta, Anthropic and several open-source players have been gaining momentum with rapid model releases and feature rollouts. Garlic appears to be OpenAI’s response: a model designed to surpass rivals on core performance where demand is highest — reasoning, stability, code generation and reliability.

If Garlic’s testing results hold true and the rollout stays on course, ChatGPT users could see a material jump in capability within the first half of next year.