Does Gemini API's Token per minute (TPM) rate limit apply only on input tokens or input + tool use tokens also?

Gemini API
1

In this documentation: https://ai.google.dev/gemini-api/docs/rate-limits it is mentioned that rate limits are also measured across input tokens per minute.

But if I add GoogleSearch or UrlContext tools, would their tokens also count in this rate limit?

Also, just to be exhaustive, output tokens would never be considered in rate limiting right?