I have been testing agents with Gemini models such as 2.0 and 2.5, however i noticed that processes and response times are extremely slow compared to other tools utilizing same model APIs or IDEs that have access to similar models.
Not sure if anyone has insight about why this might be?
Is this related to be on a trial tier vs paid tier ?
They are extremely rate limited if you are using the free versions. Its highly likely that is what you are encountering.
So its an n8n rate limiting issue possibly given its a free version.
n8n doesnt rate limit, the free gemini models do.
I am not using a free Gemini model, its a paid access through Google Cloud and Gemini APIs.
I am testing the same API through other services and responses are much much faster.
My assumption is that the nodes on trial tier for n8n are limited in resources.
I thought it was just me! Today I upgraded from 1.92 to 1.100 and noticed that Gemini responses are extremely slow now. In 1.92, my agent responded in about 5 seconds (it uses 2 agents with the Think Tool), but now it takes around 14 seconds! This is definitely a bug. Just to clarify, iām not using free api either, i have a billing account activated.