hitting gemini quota limits quickly, 1m tokens / minute

Open brig-pinkfish opened this issue 2 months ago • 2 comments

Using stagehand v3 in agent mode, with caching turned on. Doing 100s of runs - few at a time. Pretty simple workflow. It's a login, click a few things. Submit a form.

The quota is: GenerateContent input token count limit per model per minute

Not adjustable past 1m / min.

Anyone have suggestions for reducing token usage? (other than caching)?

Nov 12 '25 19:11 brig-pinkfish

Hi @brig-pinkfish is this with the cua agent mode or the default DOM-based?

Nov 13 '25 16:11 miguelg719

i was not able to get cua agent working... so was normal gemini-2.5

Nov 20 '25 21:11 brig-pinkfish

@brig-pinkfish you can use there deterministic playwright agent that does not use llm calls for each step ?

Dec 06 '25 18:12 shivamgoyal6382