I too suspect that this behaviour is the result of Google enforcing strict guardrails to the model in order to protect their intellectual property. It feels bad experiencing this without any official confirmation as to why it’s actually happening.
Same here, i just burnt through ~600 ai credits on a complicated prompt, that message appeared in between EVERY step of the thinking process.
This is a call to action for Google. The message that the agent is “specifying to appropriate tools” is of 0 consequence to the user and is therefore merely a way to set tokens alight. Your users do not NEED to see that Gemini is thinking about which tools to use. It’s like someone in the kitchen informing all the other people present that they are thinking about opening a drawer and picking up a spoon, out loud, 50 times over, before ultimately opening the drawer and picking it up in about 1 second. To me that feels like a plummer that charges by the hour and sends you an invoice for the first two days where they were thinking about which tool they might use. Not forming a plan. Just thinking of the tools to make that plan. Stop this madness or risk losing many, many AI Ultra subscribers like myself to your competitors.
I am seeing the same! Horrible!
How is this not patched months later ??