scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-08-28 05:25:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 36

test(e2e): price Together cached tokens at the registry cache read rate

Browse source
This commit is contained in:
mateo-berri 2026-08-25 18:17:28 -07:00
parent a543b0348c
commit cc6b3783fe
1 changed files with 4 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

5
tests/e2e/llm_translation/test_together_ai_e2e.py
View file

@ -423,6 +423,9 @@ class TestTogetherChatCompletions:
registry: dict[str, CostMapEntry],
reasoning_tool_backend: str,
) -> None:
"""Cached prompt tokens (Together caches shared prefixes across calls) bill at
the registry's cache read rate, which is $0 until a row carries one (LIT-5972),
so the expected cost is computed from the same registry row the proxy prices with."""
model, key = _register(client, resources, reasoning_tool_backend)
result = client.proxy.transport.send(
@ -450,7 +453,7 @@ class TestTogetherChatCompletions:
cached = (usage.prompt_tokens_details.cached_tokens or 0) if usage.prompt_tokens_details else 0
expected = (
(usage.prompt_tokens - cached) * price.input_cost_per_token
+ cached * (price.cache_read_input_token_cost or price.input_cost_per_token)
+ cached * (price.cache_read_input_token_cost or 0.0)
+ usage.completion_tokens * price.output_cost_per_token
)
assert _approx_equal(header_cost, expected), (