OpenAI's latest flagship model is **o1** (formerly known by its research codename "Strawberry").

Released in September 2024, the **o1 series** represents a new class of models designed for "reasoning." Unlike previous models (such as GPT-4o), which respond to prompts almost immediately, o1 is trained to "think" through a chain of thought before providing an answer. This makes it particularly powerful at complex tasks involving advanced mathematics, coding, and scientific problem-solving.

There are currently two versions of this series:
1.  **o1-preview:** The main high-reasoning model.
2.  **o1-mini:** A faster, more cost-effective version optimized for coding and STEM tasks.

**Note on GPT-4o:** While **GPT-4o** ("o" for omni) remains OpenAI’s flagship *multimodal* model—responsible for real-time voice, vision, and text interaction—**o1** is considered the current state-of-the-art for high-level reasoning and complex logic.
