The Assistant API is powered by Code Interpreter, a tool that writes and runs Python code, and can generate graphs, charts, and process files. The API can also access a retrieval component that supplements dev-created assistants with knowledge from outside OpenAI’s models. It also supports function calling, allowing assistants to invoke programming functions defined by a developer. The Assistant API is now in beta and available to all developers. The tokens used for the API will be billed at the chosen model’s per-token rates.
Key takeaways:
- OpenAI has announced a new Assistant API at its developer conference, which is designed to help developers create "agent-like experiences" within their apps.
- The Assistant API allows customers to build an "assistant" that can use OpenAI's generative AI models and tools to perform tasks, with potential use cases including natural language-based data analysis apps, coding assistants, and AI-powered vacation planners.
- The API is powered by Code Interpreter, a tool that writes and runs Python code in a sandboxed environment, and can generate graphs, process files, and solve complex code and math problems.
- The Assistant API is currently in beta and is available to all developers, with billing based on the chosen model's per-token rates.