Kimi K2.5 - Persistent errors with model provider
I haven’t been able to complete any requests with Kimi K2.5 both thinking and non-thinking. I thought my first attempt was too spicy but another attempt with a very boring topic also failed.
Trace ID: 49303192ef042abc6bb356d682be8e85
Trace ID: 68e43974fb4c38f98dc6c6e1d98d64ac
Chrome via Helium
Log in to comment and vote
Comments7
Sapphire Crab
Sometimes it’sd once a day and sometimes it’s every message. And because T3 chat doesn’t have a good option of re-run it using same - it takers time
Mark Florkowski
am I mmisunderstanding?
Apricot Mat
Trace ID: 4c4dc83cb9b57351b16c512e462460be
Trace ID: d30e4e98bc222178c30a7075ce09d200
Bronze Crayon
update, as far as i am aware; this has been resolved 👍
Mark Florkowski
This is a new model and highly in demand, so it may be unstable for the first week. We are doing what we can to load balance across providers and keep the availability up, but there is only so much that can be done.
Teal Ox
Hi Mark, I understand the situation.
Sorry I forgot to add some relevant detail, I think most of us users are aware that T3 Chat has a ‘service degraded’ badge displaying function and therefore we would expect it to appear during periods with elevated error rates. In this case, I didn’t see the badge so I assumed it was another issue that should be reported. Thank you!
Mark Florkowski
Unfortunately in this case the upstream providers are not reporting downtime because they are intentionall rate-limiting the requests, so our automated system did not pick up on the degraded status.