OpenAI's o1 model, internally codenamed "Project Strawberry," represents a major advancement in AI reasoning capabilities. This series of models, including o1-preview and the smaller, more affordable o1-mini, focuses on improving logical thinking, overcoming flaws like the quirky issue LLMs had with counting letters in words such as "strawberry." These models tackle the problem of hallucinations and incorrect breakdowns of tasks, particularly in how they tokenize text.

What makes o1 especially exciting is its potential to reason more like humans, handling complex tasks such as advanced mathematics, competitive programming, and even long-horizon tasks (LHTs) where the AI must plan and execute a series of steps autonomously. This reasoning breakthrough could unlock a new level of AI performance, moving beyond mere conversation into tasks requiring logical deduction and multi-step problem solving.

Currently available to ChatGPT Plus subscribers, these models offer a preview of the future of AI that not only understands language but can think critically, performing deeper, more accurate research and decision-making processes.