scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-06 08:16:43 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 79
litellm/litellm/caching
History
michelligabriele 0dd64baa66
fix(caching): preserve prompt_tokens_details through embedding cache round-trip (#26653) 
* fix(caching): preserve prompt_tokens_details through embedding cache round-trip

The embedding caching layer was dropping prompt_tokens_details (including
image_count) because CachedEmbedding had no field for usage metadata and
the cache retrieval code reconstructed Usage without it. This caused
inconsistent responses where the first call returned image_count but
cached responses did not, breaking cost tracking for multimodal embeddings.

Add prompt_tokens_details to CachedEmbedding, persist per-item details
during cache storage, aggregate them on retrieval, and merge them in
combine_usage() for partial cache hits.

* style: apply Black formatting to caching files

* fix(caching): address Greptile review — cyclic import, guarded construction, nested dict merge

Move PromptTokensDetailsWrapper to inline import to resolve CodeQL cyclic
import warning. Guard PromptTokensDetailsWrapper construction with
try/except to handle unexpected cached keys. Add recursive dict merging
in _merge_prompt_tokens_details for nested fields like
cache_creation_token_details.
2026-04-28 08:25:11 -07:00
..
__init__.py Add GCS bucket caching support (#13122) 2025-08-04 16:09:33 -07:00
_internal_lru_cache.py (litellm SDK perf improvements) - handle cases when unable to lookup model in model cost map (#7750) 2025-01-13 19:58:46 -08:00
azure_blob_cache.py style: run black formatter on entire codebase 2026-03-11 17:07:57 -03:00
base_cache.py style: run black formatter on entire codebase 2026-03-11 17:07:57 -03:00
caching.py fix(caching): preserve prompt_tokens_details through embedding cache round-trip (#26653) 2026-04-28 08:25:11 -07:00
caching_handler.py fix(caching): preserve prompt_tokens_details through embedding cache round-trip (#26653) 2026-04-28 08:25:11 -07:00
disk_cache.py [Bug Fix] No module named 'diskcache' (#11600) 2025-06-10 14:54:11 -07:00
dual_cache.py Litellm ishaan march 20 (#24303) 2026-03-21 12:40:11 -07:00
gcs_cache.py style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
in_memory_cache.py fix: prune expired in-memory cache heap entries (#25664) 2026-04-14 23:37:49 +05:30
llm_caching_handler.py fix: don't close HTTP/SDK clients on LLMClientCache eviction (#22925) 2026-03-05 12:00:38 -08:00
qdrant_semantic_cache.py style: run black formatter on entire codebase 2026-03-11 17:07:57 -03:00
Readme.md add azure blob cache support (#12587) 2025-07-15 11:47:38 -07:00
redis_cache.py Litellm ishaan march 20 (#24303) 2026-03-21 12:40:11 -07:00
redis_cluster_cache.py style: run black formatter on entire codebase 2026-03-11 17:07:57 -03:00
redis_semantic_cache.py build(pyproject.toml): add new dev dependencies - for type checking (#9631) 2025-03-29 11:02:13 -07:00
s3_cache.py style: run black formatter on entire codebase 2026-03-11 17:07:57 -03:00

Readme.md

Caching on LiteLLM

LiteLLM supports multiple caching mechanisms. This allows users to choose the most suitable caching solution for their use case.

The following caching mechanisms are supported:

  1. RedisCache
  2. RedisSemanticCache
  3. QdrantSemanticCache
  4. InMemoryCache
  5. DiskCache
  6. S3Cache
  7. AzureBlobCache
  8. DualCache (updates both Redis and an in-memory cache simultaneously)

Folder Structure

litellm/caching/
├── base_cache.py
├── caching.py
├── caching_handler.py
├── disk_cache.py
├── dual_cache.py
├── in_memory_cache.py
├── qdrant_semantic_cache.py
├── redis_cache.py
├── redis_semantic_cache.py
├── s3_cache.py

Documentation