Title: Grok 4.1 Fast is not respecting the "Thinking" Toggle
Description:
Hi support team,
I've noticed a regression in the "Thinking" vs "Instant" toggle specifically for the Grok 4.1 Fast model in t3-chat. Previously, this worked as expected across models:
Instant mode: Immediate response, no visible thinking buffer/streaming.
Thinking mode: Thinking steps shown first, then final response.
Now, even with "Instant" toggled on, there's always a thinking buffer delay before the response appears, making the toggle ineffective.
Steps to Reproduce:
Log into t3-chat.
Select Grok 4.1 Fast model.
Toggle to "Instant" mode (confirm it's active).
Send any simple query (e.g., "Hello").
Observe: Thinking buffer/stream appears before final response.
Actual Behavior:
Both modes show thinking buffer, defeating the purpose of Instant.
Environment:
Browser: Zen (Firefox based && latest version)
OS: Fedora Linux (latest)
Model: Grok 4.1 Fast only (other models work fine)
This started recently (post-update?). No huge blocker, but toggle feels broken. Happy to provide screenshots, video, or test further.
Thanks!
~ Sam
Please authenticate to join the conversation.
Completed
Feature Request
Get notified by email when there are changes.
Completed
Feature Request
Get notified by email when there are changes.