tiennm99/litellm
1
0
Fork 0
mirror of https://github.com/tiennm99/litellm.git synced 2026-08-05 20:23:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert "Fix audio cost per second override (#19158)"

Browse Source 
This reverts commit 2a0f87bde0.
This commit is contained in:
Sameer Kankute
2026-01-19 18:51:08 +05:30
committed by GitHub
parent 6eb3f579d7
commit 574391c118
3 changed files with 14 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/my-website/docs/sdk_custom_pricing.md
+1 -4
View File
@@ -2,10 +2,7 @@
Register custom pricing for sagemaker completion model.
For cost per second pricing, register `input_cost_per_second`. If your provider
charges for audio output duration (e.g., TTS), also set `output_cost_per_second`.
Values of `0` are treated as not billable, so `output_cost_per_second: 0` will
not override `input_cost_per_second`.
For cost per second pricing, you **just** need to register `input_cost_per_second`.
```python
# !pip install boto3
litellm/llms/openai/cost_calculation.py
+13 -9
View File
@@ -105,21 +105,25 @@ def cost_per_second(
prompt_cost = 0.0
completion_cost = 0.0
## Speech / Audio cost calculation
output_cost_per_second = model_info.get("output_cost_per_second")
input_cost_per_second = model_info.get("input_cost_per_second")
if output_cost_per_second is not None and output_cost_per_second > 0:
if (
"output_cost_per_second" in model_info
and model_info["output_cost_per_second"] is not None
):
verbose_logger.debug(
f"For model={model} - output_cost_per_second: {output_cost_per_second}; duration: {duration}"
f"For model={model} - output_cost_per_second: {model_info.get('output_cost_per_second')}; duration: {duration}"
)
## COST PER SECOND ##
completion_cost = output_cost_per_second * duration
if input_cost_per_second is not None and input_cost_per_second > 0:
completion_cost = model_info["output_cost_per_second"] * duration
elif (
"input_cost_per_second" in model_info
and model_info["input_cost_per_second"] is not None
):
verbose_logger.debug(
f"For model={model} - input_cost_per_second: {input_cost_per_second}; duration: {duration}"
f"For model={model} - input_cost_per_second: {model_info.get('input_cost_per_second')}; duration: {duration}"
)
## COST PER SECOND ##
prompt_cost = input_cost_per_second * duration
prompt_cost = model_info["input_cost_per_second"] * duration
completion_cost = 0.0
return prompt_cost, completion_cost
tests/test_litellm/test_cost_calculator.py
-34
View File
@@ -192,40 +192,6 @@ def test_transcription_cost_falls_back_to_duration():
assert pytest.approx(cost, rel=1e-6) == expected_cost
def test_transcription_cost_prefers_input_when_output_zero(monkeypatch):
from litellm import completion_cost
os.environ["LITELLM_LOCAL_MODEL_COST_MAP"] = "True"
litellm.model_cost = litellm.get_model_cost_map(url="")
model_name = "custom-whisper-input-only"
custom_model_info = {
"input_cost_per_second": 0.00005,
"output_cost_per_second": 0.0,
"litellm_provider": "openai",
"mode": "audio_transcription",
"supported_endpoints": ["/v1/audio/transcriptions"],
}
monkeypatch.setattr(
litellm,
"model_cost",
{**litellm.model_cost, model_name: custom_model_info},
)
response = TranscriptionResponse(text="demo text")
response.duration = 300.0
cost = completion_cost(
completion_response=response,
model=model_name,
custom_llm_provider="openai",
call_type="atranscription",
)
expected_cost = 300.0 * 0.00005
assert pytest.approx(cost, rel=1e-6) == expected_cost
def test_handle_realtime_stream_cost_calculation():
from litellm.cost_calculator import RealtimeAPITokenUsageProcessor