ggerganov/llama.cpp Released Version b12431
llama.cpp release b12431 improves server error handling by returning a 400 status code when unsupported reasoning effort levels are requested. Instead of throwing a server 500 error, requests with invalid effort levels are now correctly rejected during parameter parsing. This update improves API robustness and compliance for developers integrating local LLM servers with OpenAI-compatible endpoints. Proper error codes make debugging client applications significantly smoother. The update modifies oaicompat_chat_params_parse to catch unsupported reasoning effort parameters and list the supported levels in the response. Valid aliases and levels ignored by the template remain unaffected.
## BACKGROUND
llama.cpp is an open-source library that enables efficient local inference for large language models using the GGML tensor library. Its built-in server provides an OpenAI-compatible API interface, allowing developers to run models locally while using standard API structures.