KamilKurde Assistant barely works. This simple question took 40 seconds for grok 4 to respond (which is expected since it's a preview) https://kagi.com/assistant/fd270a49-be0e-4bc1-9db4-10a4c2ba9f17 But other models have trouble producing answers too. Deepseek R1 took A HUNDRED SECONDS to list first 10 prime numbers https://kagi.com/assistant/cc44ac06-572d-46a3-b665-6a3b48c9ba09 and it wasn't that slow before. It got to the point that some queries won't be answered at all because before the answer is finished I get a Coś poszło nie tak podczas uzyskiwania twojej odpowiedzi TypeError: network error, and clicking on retry resets the chat without producing any output I expect assistant to works correctly, as according to the sidebar "All services are working" (EUROPE-WEST2) but that's evidently not the case
laiz Hey! Another user mentioned a similar issue on discord where they were only able to get ~ 2 tokens per second. Oddly, however, this doesn't seem to affect all of us on EU-WEST2
kirkmc I've had problems today with both the assistant and quick answers. I'm in the UK if it matters.
laiz @KamilKurde @kirkmc others are reporting that the speed is back to what they'd expect — could you check if that is the case for you as well?
KamilKurde laiz It seems to be alright, but KI still seems to throw TypeError: network error on first response
KamilKurde yiwei-1 Example thread in which it happens: https://kagi.com/assistant/feb7898c-be14-4b3b-83bb-dd061fdbcfee Here, it's on second branch (because first branch worked correctly) but I recall yesterday (when I was testing, sorry but I've deleted some of the threads I was testing shit on) I've seen it even on first branch, on first response, in desktop Vivaldi browser
