What is the Gemini context window and how do you manage long contexts?

from google import genai
from google.genai import types

client = genai.Client()

# Count tokens before sending (avoid context overflow errors):
token_response = client.models.count_tokens(
    model="gemini-3.5-flash",
    contents="Your very long document here...",
)
print(f"Token count: {token_response.total_tokens}")  # check vs 1M limit

# Long context use cases Gemini excels at:
# 1. Entire codebase analysis
response = client.models.generate_content(
    model="gemini-3.5-flash",
    contents=[
        "Summarise all TODO comments in this codebase and prioritise them.",
        entire_codebase_text,  # can be hundreds of files
    ]
)

# 2. Full book Q&A
# 3. Hour-long video analysis
# 4. Full conversation history

# Context caching: cache large stable prompts to reduce cost:
cached = client.caches.create(
    model="gemini-3.5-flash",
    contents=[large_document],  # cache this expensive prefix
    ttl="3600s",
)
# Reference the cache in subsequent calls:
response = client.models.generate_content(
    model="gemini-3.5-flash",
    contents="Summarise section 3.",
    config=types.GenerateContentConfig(cached_content=cached.name),
)