scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-11 22:51:28 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 123

Update utils.py

Browse source
This commit is contained in:
Praveen11558 2026-03-24 23:01:58 +05:30 • committed by GitHub
parent 238891d169
commit 64161bc85d
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 11 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
litellm/litellm_core_utils/llm_cost_calc/utils.py
View file

@ -673,13 +673,17 @@ def generic_cost_per_token( # noqa: PLR0915
)
has_double_counting = cache_hit > 0 and accounted_tokens > usage.prompt_tokens
# Some models use alternative billing dimensions (character_count, video_length_seconds)
# that account for prompt_tokens without being included in the per-token detail fields.
# When these are active, a gap between accounted_tokens and prompt_tokens is expected
# and must NOT be filled. Note: image_count is additive billing, not alternative.
# Some models use alternative billing dimensions that account for prompt_tokens
# without being included in the per-token detail fields. When any of these are
# active, a gap between accounted_tokens and prompt_tokens is expected and must
# NOT be filled:
# - character_count / video_length_seconds: replace per-token billing entirely
# - image_count: additive flat-rate billing; the image-URL tokens are already
# reflected in prompt_tokens and billed separately via input_cost_per_image
has_non_token_alternative_billing = (
prompt_tokens_details["character_count"] > 0
or prompt_tokens_details["video_length_seconds"] > 0
or prompt_tokens_details["image_count"] > 0
)
if has_double_counting:
@ -697,7 +701,6 @@ def generic_cost_per_token( # noqa: PLR0915
elif (
accounted_tokens < usage.prompt_tokens
and not has_non_token_alternative_billing
and prompt_tokens_details["image_count"] == 0
):
# Unaccounted tokens fix: inline documents (PDF, DOCX, etc.) are counted
# in prompt_tokens by the provider but not broken out into any detail field.
@ -707,10 +710,9 @@ def generic_cost_per_token( # noqa: PLR0915
# 1. text_tokens=0 (provider didn't set it) - gap fills the entire remainder
# 2. text_tokens>0 but < total (mixed content like PDF+text) - gap fills remainder
#
# Skip filling when:
# - Non-token alternative billing is active (character_count, video_length_seconds)
# - Image count billing is active (image_count > 0) - those tokens are already
# billed separately via input_cost_per_image and shouldn't be double-charged
# Skip filling when has_alternative_billing is True (character_count,
# video_length_seconds, or image_count are active) - those dimensions
# already account for the prompt_tokens gap.
unaccounted_tokens = usage.prompt_tokens - accounted_tokens
prompt_tokens_details["text_tokens"] += unaccounted_tokens