Access the LLM API of different providers using a unified SDK.
- Supports multiple LLM providers: OpenAI, Anthropic, Google Gemini, Cohere, Mistral
- Supports multi modalities (text, image, audio), including OpenAI newest audio generation capabilities.
- Supports reporting the token usage and calculating cost of a request if given the model's pricing information.
The specification serves as the basis to implement the unified LLM SDK in different programming languages. The specification is expressed using JSON schema and can be found in schema.json.
Implementations in different programming languages should adhere to this specification but may adapt it to the idioms of the respective language or provide additional functionalities.