phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-27 11:43:54 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

rename llms/OpenAI/ -> llms/openai/ (#7154)

Browse source 
* rename OpenAI -> openai

* fix file rename

* fix rename changes

* fix organization of openai/transcription

* fix import OA fine tuning API

* fix openai ft handler

* fix handler import
This commit is contained in:
Ishaan Jaff 2024-12-10 20:14:07 -08:00 • committed by GitHub
parent 61afdab228
commit 5ad57dd54b
48 changed files with 53 additions and 59 deletions
Download patch file Download diff file Expand all files Collapse all files

113
litellm/llms/OpenAI/cost_calculation.py
View file

@ -1,113 +0,0 @@
"""
Helper util for handling openai-specific cost calculation
- e.g.: prompt caching
"""
from typing import Literal, Optional, Tuple
from litellm._logging import verbose_logger
from litellm.types.utils import CallTypes, Usage
from litellm.utils import get_model_info
def cost_router(call_type: CallTypes) -> Literal["cost_per_token", "cost_per_second"]:
if call_type == CallTypes.atranscription or call_type == CallTypes.transcription:
return "cost_per_second"
else:
return "cost_per_token"
def cost_per_token(model: str, usage: Usage) -> Tuple[float, float]:
"""
Calculates the cost per token for a given model, prompt tokens, and completion tokens.
Input:
- model: str, the model name without provider prefix
- usage: LiteLLM Usage block, containing anthropic caching information
Returns:
Tuple[float, float] - prompt_cost_in_usd, completion_cost_in_usd
"""
## GET MODEL INFO
model_info = get_model_info(model=model, custom_llm_provider="openai")
## CALCULATE INPUT COST
### Non-cached text tokens
non_cached_text_tokens = usage.prompt_tokens
cached_tokens: Optional[int] = None
if usage.prompt_tokens_details and usage.prompt_tokens_details.cached_tokens:
cached_tokens = usage.prompt_tokens_details.cached_tokens
non_cached_text_tokens = non_cached_text_tokens - cached_tokens
prompt_cost: float = non_cached_text_tokens * model_info["input_cost_per_token"]
## Prompt Caching cost calculation
if model_info.get("cache_read_input_token_cost") is not None and cached_tokens:
# Note: We read ._cache_read_input_tokens from the Usage - since cost_calculator.py standardizes the cache read tokens on usage._cache_read_input_tokens
prompt_cost += cached_tokens * (
model_info.get("cache_read_input_token_cost", 0) or 0
)
_audio_tokens: Optional[int] = (
usage.prompt_tokens_details.audio_tokens
if usage.prompt_tokens_details is not None
else None
)
_audio_cost_per_token: Optional[float] = model_info.get(
"input_cost_per_audio_token"
)
if _audio_tokens is not None and _audio_cost_per_token is not None:
audio_cost: float = _audio_tokens * _audio_cost_per_token
prompt_cost += audio_cost
## CALCULATE OUTPUT COST
completion_cost: float = (
usage["completion_tokens"] * model_info["output_cost_per_token"]
)
_output_cost_per_audio_token: Optional[float] = model_info.get(
"output_cost_per_audio_token"
)
_output_audio_tokens: Optional[int] = (
usage.completion_tokens_details.audio_tokens
if usage.completion_tokens_details is not None
else None
)
if _output_cost_per_audio_token is not None and _output_audio_tokens is not None:
audio_cost = _output_audio_tokens * _output_cost_per_audio_token
completion_cost += audio_cost
return prompt_cost, completion_cost
def cost_per_second(
model: str, usage: Usage, response_time_ms: Optional[float] = 0.0
) -> Tuple[float, float]:
"""
Calculates the cost per second for a given model, prompt tokens, and completion tokens.
"""
## GET MODEL INFO
model_info = get_model_info(model=model, custom_llm_provider="openai")
prompt_cost = 0.0
completion_cost = 0.0
## Speech / Audio cost calculation
if (
"output_cost_per_second" in model_info
and model_info["output_cost_per_second"] is not None
and response_time_ms is not None
):
verbose_logger.debug(
f"For model={model} - output_cost_per_second: {model_info.get('output_cost_per_second')}; response time: {response_time_ms}"
)
## COST PER SECOND ##
completion_cost = model_info["output_cost_per_second"] * response_time_ms / 1000
elif (
"input_cost_per_second" in model_info
and model_info["input_cost_per_second"] is not None
and response_time_ms is not None
):
verbose_logger.debug(
f"For model={model} - input_cost_per_second: {model_info.get('input_cost_per_second')}; response time: {response_time_ms}"
)
## COST PER SECOND ##
prompt_cost = model_info["input_cost_per_second"] * response_time_ms / 1000
completion_cost = 0.0
return prompt_cost, completion_cost