scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-21 00:21:49 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 74

fix(cost): fill the missing realtime cache-read rates

Browse source
This commit is contained in:
mateo-berri 2026-09-12 17:48:00 -07:00
parent 4b84c83788
commit a94c060b84
3 changed files with 42 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
litellm/model_prices_and_context_window_backup.json
View file

@ -5513,7 +5513,8 @@
},
"azure/gpt-realtime-2025-08-28": {
"cache_creation_input_audio_token_cost": 4e-06,
"cache_read_input_token_cost": 4e-06,
"cache_read_input_audio_token_cost": 4e-07,
"cache_read_input_token_cost": 4e-07,
"deprecation_date": "2027-03-02",
"input_cost_per_audio_token": 3.2e-05,
"input_cost_per_image_token": 5e-06,
@ -5546,7 +5547,8 @@
},
"azure/gpt-realtime-1.5-2026-02-23": {
"cache_creation_input_audio_token_cost": 4e-06,
"cache_read_input_token_cost": 4e-06,
"cache_read_input_audio_token_cost": 4e-07,
"cache_read_input_token_cost": 4e-07,
"deprecation_date": "2027-08-24",
"input_cost_per_audio_token": 3.2e-05,
"input_cost_per_image_token": 5e-06,
@ -5683,6 +5685,7 @@
},
"azure/gpt-realtime-mini": {
"cache_creation_input_audio_token_cost": 3e-07,
"cache_read_input_audio_token_cost": 3e-07,
"cache_read_input_token_cost": 6e-08,
"input_cost_per_audio_token": 1e-05,
"input_cost_per_image_token": 8e-07,
@ -5715,6 +5718,7 @@
},
"azure/gpt-realtime-mini-2025-10-06": {
"cache_creation_input_audio_token_cost": 3e-07,
"cache_read_input_audio_token_cost": 3e-07,
"cache_read_input_token_cost": 6e-08,
"input_cost_per_audio_token": 1e-05,
"input_cost_per_image_token": 8e-07,
@ -32683,6 +32687,7 @@
"gpt-realtime-mini": {
"cache_creation_input_audio_token_cost": 3e-07,
"cache_read_input_audio_token_cost": 3e-07,
"cache_read_input_token_cost": 6e-08,
"deprecation_date": "2027-01-20",
"input_cost_per_audio_token": 1e-05,
"input_cost_per_token": 6e-07,

9
model_prices_and_context_window.json
View file

@ -5513,7 +5513,8 @@
},
"azure/gpt-realtime-2025-08-28": {
"cache_creation_input_audio_token_cost": 4e-06,
"cache_read_input_token_cost": 4e-06,
"cache_read_input_audio_token_cost": 4e-07,
"cache_read_input_token_cost": 4e-07,
"deprecation_date": "2027-03-02",
"input_cost_per_audio_token": 3.2e-05,
"input_cost_per_image_token": 5e-06,
@ -5546,7 +5547,8 @@
},
"azure/gpt-realtime-1.5-2026-02-23": {
"cache_creation_input_audio_token_cost": 4e-06,
"cache_read_input_token_cost": 4e-06,
"cache_read_input_audio_token_cost": 4e-07,
"cache_read_input_token_cost": 4e-07,
"deprecation_date": "2027-08-24",
"input_cost_per_audio_token": 3.2e-05,
"input_cost_per_image_token": 5e-06,
@ -5683,6 +5685,7 @@
},
"azure/gpt-realtime-mini": {
"cache_creation_input_audio_token_cost": 3e-07,
"cache_read_input_audio_token_cost": 3e-07,
"cache_read_input_token_cost": 6e-08,
"input_cost_per_audio_token": 1e-05,
"input_cost_per_image_token": 8e-07,
@ -5715,6 +5718,7 @@
},
"azure/gpt-realtime-mini-2025-10-06": {
"cache_creation_input_audio_token_cost": 3e-07,
"cache_read_input_audio_token_cost": 3e-07,
"cache_read_input_token_cost": 6e-08,
"input_cost_per_audio_token": 1e-05,
"input_cost_per_image_token": 8e-07,
@ -32683,6 +32687,7 @@
"gpt-realtime-mini": {
"cache_creation_input_audio_token_cost": 3e-07,
"cache_read_input_audio_token_cost": 3e-07,
"cache_read_input_token_cost": 6e-08,
"deprecation_date": "2027-01-20",
"input_cost_per_audio_token": 1e-05,
"input_cost_per_token": 6e-07,

28
tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
View file

@ -5275,3 +5275,31 @@ def test_cache_read_breakdown_splits_cached_audio_at_the_audio_cache_rate(_local
assert breakdown.rates is not None
assert breakdown.rates.cache_read_input_audio_token_cost == pytest.approx(3e-7)
assert prompt_cost == pytest.approx((1693 - 896) * 6e-7 + (3170 - 1920) * 1e-5 + breakdown.cache_read_cost)
@pytest.mark.parametrize(
("model", "custom_llm_provider", "expected_prompt_cost"),
(
pytest.param("azure/gpt-realtime-2025-08-28", "azure", 300 * 4e-6 + 100 * 4e-7 + 200 * 3.2e-5 + 400 * 4e-7, id="azure-gpt-realtime"),
pytest.param("azure/gpt-realtime-1.5-2026-02-23", "azure", 300 * 4e-6 + 100 * 4e-7 + 200 * 3.2e-5 + 400 * 4e-7, id="azure-gpt-realtime-1.5"),
pytest.param("azure/gpt-realtime-mini", "azure", 300 * 6e-7 + 100 * 6e-8 + 200 * 1e-5 + 400 * 3e-7, id="azure-gpt-realtime-mini"),
pytest.param("gpt-realtime-mini", "openai", 300 * 6e-7 + 100 * 6e-8 + 200 * 1e-5 + 400 * 3e-7, id="openai-gpt-realtime-mini"),
),
)
def test_realtime_models_bill_cached_text_and_audio_at_their_cache_read_rates(
_local_model_cost_map: None, model: str, custom_llm_provider: str, expected_prompt_cost: float
) -> None:
usage = Usage(
prompt_tokens=1000,
completion_tokens=0,
total_tokens=1000,
prompt_tokens_details=PromptTokensDetailsWrapper(
text_tokens=400,
audio_tokens=600,
cached_tokens=500,
cached_tokens_details={"text_tokens": 100, "audio_tokens": 400},
),
)
prompt_cost, _ = generic_cost_per_token(model=model, usage=usage, custom_llm_provider=custom_llm_provider)
assert prompt_cost == pytest.approx(expected_prompt_cost)