Hosted Cache - api.litellm.ai (removed)

The hosted cache backed by api.litellm.ai has been removed from LiteLLM. "hosted" is not a valid Cache(type=...) value, and passing it leaves the cache without a backend, so the next completion() call raises AttributeError: 'Cache' object has no attribute 'cache'

Use one of the supported backends instead: local (the default, in memory), redis, redis-semantic, valkey-semantic, qdrant-semantic, s3, gcs, azure-blob or disk. See Caching - In-Memory, Redis, s3, gcs, Redis Semantic Cache, Disk for setup of each one. The examples below use the default in-memory cache

Quick Start Usage - Completion

import litellm
from litellm import completion
from litellm.caching.caching import Cache
litellm.cache = Cache() # in-memory cache

# Make completion calls
response1 = completion(
    model="gpt-5.6-luna", 
    messages=[{"role": "user", "content": "Tell me a joke."}],
    caching=True
)

response2 = completion(
    model="gpt-5.6-luna", 
    messages=[{"role": "user", "content": "Tell me a joke."}],
    caching=True
)
# response1 == response2, response 1 is cached

Usage - Embedding()

import time
import litellm
from litellm import completion, embedding
from litellm.caching.caching import Cache
litellm.cache = Cache()

start_time = time.time()
embedding1 = embedding(model="text-embedding-ada-002", input=["hello from litellm"*5], caching=True)
end_time = time.time()
print(f"Embedding 1 response time: {end_time - start_time} seconds")

start_time = time.time()
embedding2 = embedding(model="text-embedding-ada-002", input=["hello from litellm"*5], caching=True)
end_time = time.time()
print(f"Embedding 2 response time: {end_time - start_time} seconds")

Caching with Streaming

LiteLLM can cache your streamed responses for you

Usage

import litellm
import time
from litellm import completion
from litellm.caching.caching import Cache

litellm.cache = Cache()

# Make completion calls
response1 = completion(
    model="gpt-5.6-luna", 
    messages=[{"role": "user", "content": "Tell me a joke."}], 
    stream=True,
    caching=True)
for chunk in response1:
    print(chunk)

time.sleep(1) # cache is updated asynchronously

response2 = completion(
    model="gpt-5.6-luna", 
    messages=[{"role": "user", "content": "Tell me a joke."}], 
    stream=True,
    caching=True)
for chunk in response2:
    print(chunk)