Optional
fields: ChatCohereInputThe name of the model to use.
{"command"}
Whether or not to include token usage when streaming.
This will include an extra chunk at the end of the stream
with eventType: "stream-end"
and the token usage in
usage_metadata
.
{true}
Whether or not to stream the response.
{false}
What sampling temperature to use, between 0.0 and 2.0. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.
{0.3}
Generated using TypeDoc
Integration with ChatCohere
Example