OpenAI Unveils Faster GPT-4 Turbo and Assistants API at Developer Day
OpenAI introduces a more capable GPT-4 Turbo model with a 128K context window alongside a new Assistants API to simplify building AI agents. The company also significantly reduces pricing across its platform.
OpenAI unveils its new GPT-4 Turbo model at its first-ever Developer Day conference, bringing significant improvements over the original GPT-4. This upgraded model supports a 128K context window, allowing it to process over 300 pages of text in a single prompt, and includes knowledge of world events up to April 2023. Alongside the new model, OpenAI announces substantial price reductions, making GPT-4 Turbo input tokens three times cheaper and output tokens two times cheaper than the previous version.
The company also introduces the new Assistants API, which is currently in beta and available to all developers. This API allows developers to easily build agent-like experiences directly into their applications by creating purpose-built AI entities with specific instructions. It includes powerful new tools like Code Interpreter and Retrieval to handle complex tasks that previously required extensive custom coding from developers.
Additionally, OpenAI updates its function calling capabilities to make integrations much more seamless for developers. The improved feature allows the model to call multiple functions in a single message and outputs a JSON object with the correct arguments for external APIs. Microsoft CEO Satya Nadella highlights these advancements, noting that Azure is drastically changing its infrastructure to support OpenAI's APIs and that Microsoft is building its own Copilot features as a developer on the OpenAI platform.