January 2026 (2026-01) monthly summary for ggml-org/llama.cpp: Implemented structured reasoning content blocks support for the Anthropic Messages API, enabling reasoning content to be returned alongside text for both non-streaming and streaming scenarios. This included proper indexing and state management for reasoning content, with refactoring to improve clarity and safety. Addressed streaming correctness for thinking content blocks by adding signature field and fixing duplicate content_block_start events. Replaced unsafe raw pointers in streaming state with safer field-copy patterns, improving stability and maintainability. Validated changes against bartowski/DeepSeek-R1-Distill-Qwen-7B-GGUF model and prepared for broader deployment. These changes advance model reasoning workflows, improve API usefulness, and strengthen code quality and test coverage.