[Bug] api issue

when I work with api with models I have a model then normally responds then can send that the generation is going to my agent and API monitor says that the generation is going but even after 30 minutes the generation does not start that is the model in principle did not give any commands and when you send the message again everything works but not for a long time in the console there are no errors. I use the latest version of unsloth and this error was in my cline, hermes, opencode. it didn't matter if I chose a small model with a speed of 100 t/s or a larger one, it still froze. Please help me fix this