CrewAI crashing with local llama.cpp

CrewAI Community Support
1

Hello there

I’m trying to use a local llama.cpp server with CrewAI, but I keep getting the following error from the llama server, which also causes the CrewAI code to crash:

got exception: {"code":500,"message":"Cannot have 2 or more assistant messages at the end of the list.","type":"server_error"}
srv  log_server_r: request: POST /v1/chat/completions 127.0.0.1 500

I’ve tried to host gemma3, deepseek and qwen. and got the same issue.

Is there a solution to this issue? I also tried using Ollama, but encountered a different problem, possibly related to LiteLLM.

Thanks!