Annyoing cut-offs
I love the idea of T3. Nice to have one place for keeping the chat history, regardless of which LLM you are working wit.
Lately there have been more and more glitches, i.e. cut-offs in the response stream. At this point it’s simply too much and T3 is currently not usable. Out of the last 10 responses from Geminiwithin my current discussion, half of them was cut off. Some was cut of after less than 30 lines.
I don’t know if this if a T3 or a Gemini 2.5 problem. I do know that 8 USD is a good price for having access to several models. But 8 bucks is too much if it’s not usable.
Is this a known problem with either T3 och Gemini?
Gemini 2.5 pro is currently the best model IMO. I’m thinking of moving to a Gemini subscription, but I’d loose the flexibility with T3.
Log in to comment and vote
Comments5
Apricot Gizmo
Thanks for your reply.
I tried O4 mini and Gemini Flash as well. After a couple of tries, Flash responded. Nothing works great for the moment. Maybe this is my que to call it a day. Will try again tomorrow.
I assume the models available in T3 is not top tier regarding response time etc? Would I be better off subscribing to Gemini Advanced directly?
Apricot Gizmo
@Mark Florkowski Sorry for posting this reply wrongly, should have klicked reply under your comment instead.
Mark Florkowski
No worries
Mark Florkowski
They should be the same experience, however Google may prioritize their own application over 3rd parties using the API, especially during times of high load.
I am investigating issues with reasoning models overall, as they do seem to have a higher than normal failure rate right now, but have not yet found a root cause.
Mark Florkowski
The gemini 2.5 API has been having pretty significant issues for the past week. Unfortunately we are at the mercy of Google here on how quickly they can resolve the issues on their end.