4 ms·
I also believe some providers fallback to another model entirely. I was recently using Kimi K3 and saw that some requests had no reasoning trace whatsoever. Uns
by jwxz 7d ago
I also believe some providers fallback to another model entirely. I was recently using Kimi K3 and saw that some requests had no reasoning trace whatsoever. Unsurprisingly, those requests were routed to the less reputable providers (Sail Research).
- neilmovva 7d agohi, I'm one of the founders of Sail. I'm very sorry that you had a bad experience with us! We are serious about serving models correctly, and always publish a link to the exact HF checkpoint we're using for each model in our docs. If you ever have an issue like this again, please send a note to support@sailresearch.com and we'll make it right with a detailed postmortem.
- jwxz 7d agoHey, thanks for reaching out. Just wanted to share my experience. I really thought there was an issue in your API. I had quite a few Kimi K3 requests served by Sail. Most of these were fine and had the reasoning traces (I like reading them), however I noticed that some requests were being generated unusually quick and had no reasoning trace, which led me to believe another fallback model was being used, since Kimi K3 always emits reasoning tokens.