Hi everyone,
I’m opening this discussion to see if others are experiencing significant performance and quota issues with the new Gemini 3.1 Pro Preview.
Lately, I’ve noticed:
-
Infinite Thinking Loops: The model stays in “thinking” mode for over 60-90 seconds even for relatively straightforward queries.
-
Resolution/Streaming Lag: Instead of smooth token streaming, the response “freezes” and then dumps the entire text block at once after a long delay.
-
Massive Token Consumption: The “Deep Think” process seems to be consuming tokens at an alarming rate. Even after just a few requests, I’m hitting the quota limit almost instantly.
-
24-Hour Lockout: Once the limit is reached, I’m getting blocked and told to “wait 24 hours” before I can execute anything again. This is making it impossible to use the model for professional AI development or Data Analysis workflows.
Is anyone else facing this “quota drain”? Is the internal reasoning counting against our daily token limit in a way that’s bugged or just extremely expensive?