OpenAI's new o1-pro API launched with high costs

OpenAI has launched a new API called o1-pro, which costs significantly more than previous models. Starting today, developers can access o1-pro through OpenAI’s paid tiers. The API prices are set at $150 for every 1 million input tokens and $600 for output tokens. o1-pro is designed for complex reasoning tasks. This model is one of the most expensive offerings in the market, surpassing even GPT-4.5, which costs $75 for input and $150 for output tokens. OpenAI hopes that this high price will attract researchers and professionals in fields like science and medicine. The new API provides a large context window of 200,000 tokens and allows for 100,000 max output tokens. o1-pro can handle data from both text and images, although it only produces text outputs. It also includes features like function calling and structured outputs, which ensure responses meet a developer’s specifications. Additionally, o1-pro supports integration with other APIs, allowing for more flexible use cases. However, OpenAI has not revealed the specific knowledge cutoff date for this model, though earlier versions were informed by data up until late 2023. There is increasing competition in the reasoning model space. OpenAI's o1-pro competes with models from companies like DeepSeek, Anthropic, and Google.