To the AI Studio Engineering Team:
As a developer utilizing Gemini 1.5 Pro (Latest) for high-density, iterative tasks, I am encountering critical friction with the current UI and rate-limit architecture.
- UI TOKEN COUNTER HANG: The total token counter (top left oval) frequently hangs in a perpetual loading state during high-context sessions. The async counting service is timing out, leaving users blind to their total payload weight before execution.
- REQUEST: Implement a live, localized TPM (Tokens Per Minute) counter specifically for the text input box, matching the pre-calculation behavior currently used for file/image uploads.
- COLLABORATIVE BOTTLENECK (RPM/TPM limits): The current 2-15 RPM and restrictive TPM limits severely penalize conversational, iterative workflows. When collaborating with the model on complex logic, users naturally engage in rapid-fire clarification. Because the entire context window is re-sent with every short prompt, the TPM limit trips immediately, resulting in 5+ minute server-side queue delays.
We need a mitigation for conversational ping-ponging within heavy context threads that doesn’t penalize the user with massive TPM bloat for a 2-sentence clarification.
- Crazyknome