Application developers who access OpenAI through its long-running API will now have access to the company's latest full o1 model, rather than the months-old o1-preview. The upgrade is one of a number of new features being rolled out to OpenAI's developer customers starting today.
OpenAI announced during a livestream today that the new o1 model would bring back many of the "core features you've been missing" from the API during o1-preview. This includes the ability to use developer messages to help guide your particular chatbot (i.e. "You are a helpful assistant for tax professionals.") and access to a "reasoning effort" parameter that tells the API how long to think about specific queries, saving time and money on simple problems so they can be used on the tougher ones. API users can also use visual information like scans of documents as input.
OpenAI also highlighted improvements to the new API's use of internal function calling, where the OpenAI model decides to call functions pre-written by outside developers to generate answers to certain queries, when appropriate. The new API is also more accurate in its use of structured outputs, which use a JSON schema to present information in a specific format outlined by the developer, OpenAI said.