Allow creating a new Customer
Parameters:
-
user_id: str - The unique identifier for the user.
-
alias: Optional[str] - A human-friendly alias for the user.
-
blocked: bool - Flag to allow or disallow requests for this end-user. Default is False.
-
max_budget: Optional[float] - The maximum budget allocated to the user. Either 'max_budget' or 'budget_id' should be provided, not both.
-
budget_id: Optional[str] - The identifier for an existing budget allocated to the user. Either 'max_budget' or 'budget_id' should be provided, not both.
-
allowed_model_region: Optional[Union[Literal["eu"], Literal["us"]]] - Require all user requests to use models in this specific region.
-
default_model: Optional[str] - If no equivalent model in the allowed region, default all requests to this model.
-
metadata: Optional[dict] = Metadata for customer, store information for customer. Example metadata = {"data_training_opt_out": True}
-
budget_duration: Optional[str] - Budget is reset at the end of specified duration. If not set, budget is never reset. You can set duration as seconds ("30s"), minutes ("30m"), hours ("30h"), days ("30d").
-
tpm_limit: Optional[int] - [Not Implemented Yet] Specify tpm limit for a given customer (Tokens per minute)
-
rpm_limit: Optional[int] - [Not Implemented Yet] Specify rpm limit for a given customer (Requests per minute)
-
model_max_budget: Optional[dict] - [Not Implemented Yet] Specify max budget for a given model. Example: {"openai/gpt-4o-mini": {"max_budget": 100.0, "budget_duration": "1d"}}
-
max_parallel_requests: Optional[int] - [Not Implemented Yet] Specify max parallel requests for a given customer.
-
soft_budget: Optional[float] - [Not Implemented Yet] Get alerts when customer crosses given budget, doesn't block requests.
-
Allow specifying allowed regions
-
Allow specifying default model
Example curl:
curl --location 'http://0.0.0.0:4000/customer/new' --header 'Authorization: Bearer sk-1234' --header 'Content-Type: application/json' --data '{
"user_id" : "z-jaff-3",
"allowed_region": "eu",
"budget_id": "free_tier",
"default_model": "azure/gpt-3.5-turbo-eu" <- all calls from this user, use this model?
}'
# return end-user object
NOTE: This used to be called /end_user/new
, we will still be maintaining compatibility for /end_user/XXX for these endpoints