phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-26 03:04:13 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix(utils.py): update get_model_info docstring

Browse source 
Fixes https://github.com/BerriAI/litellm/issues/4711
This commit is contained in:
Krrish Dholakia 2024-07-15 18:18:50 -07:00
parent ff1bcdd3b5
commit 9cc2daeec9
2 changed files with 48 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

15
litellm/proxy/proxy_server.py
View file

@ -5474,8 +5474,19 @@ async def new_end_user(
):
"""
Allow creating a new Customer
NOTE: This used to be called `/end_user/new`, we will still be maintaining compatibility for /end_user/XXX for these endpoints
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[Literal["eu"]] - 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}
- Allow specifying allowed regions
- Allow specifying default model
@ -5493,6 +5504,8 @@ async def new_end_user(
# 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
"""
global prisma_client, llm_router
"""

41
litellm/utils.py
View file

@ -4665,13 +4665,40 @@ def get_model_info(model: str, custom_llm_provider: Optional[str] = None) -> Mod
Returns:
dict: A dictionary containing the following information:
- max_tokens (int): The maximum number of tokens allowed for the given model.
- input_cost_per_token (float): The cost per token for input.
- output_cost_per_token (float): The cost per token for output.
- litellm_provider (str): The provider of the model (e.g., "openai").
- mode (str): The mode of the model (e.g., "chat" or "completion").
- supported_openai_params (List[str]): A list of supported OpenAI parameters for the model.
max_tokens: Required[Optional[int]]
max_input_tokens: Required[Optional[int]]
max_output_tokens: Required[Optional[int]]
input_cost_per_token: Required[float]
input_cost_per_character: Optional[float] # only for vertex ai models
input_cost_per_token_above_128k_tokens: Optional[float] # only for vertex ai models
input_cost_per_character_above_128k_tokens: Optional[
float
] # only for vertex ai models
input_cost_per_image: Optional[float] # only for vertex ai models
input_cost_per_audio_per_second: Optional[float] # only for vertex ai models
input_cost_per_video_per_second: Optional[float] # only for vertex ai models
output_cost_per_token: Required[float]
output_cost_per_character: Optional[float] # only for vertex ai models
output_cost_per_token_above_128k_tokens: Optional[
float
] # only for vertex ai models
output_cost_per_character_above_128k_tokens: Optional[
float
] # only for vertex ai models
output_cost_per_image: Optional[float]
output_vector_size: Optional[int]
output_cost_per_video_per_second: Optional[float] # only for vertex ai models
output_cost_per_audio_per_second: Optional[float] # only for vertex ai models
litellm_provider: Required[str]
mode: Required[
Literal[
"completion", "embedding", "image_generation", "chat", "audio_transcription"
]
]
supported_openai_params: Required[Optional[List[str]]]
supports_system_messages: Optional[bool]
supports_response_schema: Optional[bool]
supports_vision: Optional[bool]
Raises:
Exception: If the model is not mapped yet.