Model inference¶
Start by listing callable models to obtain an inference model identifier, then select an API supported by the model.
Choose a task¶
Task |
Page |
Description |
|---|---|---|
List callable models |
List the Genesis models available to the current credential and obtain their identifiers, capabilities, and prices. |
|
Chat generation (Chat Completions) |
Use the Chat Completions API to generate a reply from input messages. The creation endpoint describes streaming requests and event handling. |
|
Message generation (Messages) |
Use the Anthropic Messages protocol to generate a reply. Before generation, you can count the tokens in the input messages with the token-counting endpoint, which does not generate a reply. |
|
Response generation and management (Responses) |
Use the Responses API to generate content. With a personal access token and a model that supports stateful requests, you can also save generation records, then retrieve their inputs and results or delete them. |
|
Manage local conversations |
Maintain local conversation metadata and item summaries; your application maintains message bodies and model request context. |
|
Embeddings |
Convert input content into vectors. |
|
Document reranking |
Rank candidate documents by their relevance to a query. |