Can't get the implicit caching to work

Gemini API
,
9

Hi @Volodya_Bochek,
“Persistent Session” usually refers to Implicit Caching (automatic), whereas passing content once to refer to it later is Explicit Caching (manual).
“Explicit Caching” is distinct from a “Chat Session.” Explicit caching uses a cached_content ID, while a session uses history.
With the Gemini API’s explicit caching feature, you can pass large content to the model once to create a cached object. For all subsequent requests, you simply refer to those cached tokens, allowing the model to reuse the data without re-processing the entire instruction.
For Example:

from google import genai
from google.genai import types

client = genai.Client(api_key="YOUR_API_KEY")
model_name = "gemini-1.5-flash"

# 1. Create the cache (Pass content once)
cache = client.caches.create(
    model=model_name,
    config=types.CreateCachedContentConfig(
        display_name="system_instruction_cache",
        system_instruction="INSERT_YOUR_LARGE_TEXT_CONTENT_HERE",
        ttl_seconds=3600, # Cache lives for 1 hour
    )
)

# 2. Use the cache (Refer to it in subsequent requests)
response = client.models.generate_content(
    model=model_name,
    contents="Based on the cached instructions, please analyze this.",
    config=types.GenerateContentConfig(
        cached_content=cache.name  # This refers to the stored tokens
    )
)

print(f"Tokens reused from cache: {response.usage_metadata.cached_content_token_count}")