Show HN: llamafile 0.10.0 rebuilt, Qwen3.5, lfm2, Anthropic API (blog.mozilla.ai)

The AI community welcomes the release of llamafile 0.10.0, a significant overhaul that enhances portability and compatibility with upstream dependencies. This new version integrates a polyglot build of llama.cpp, allowing seamless access to both core features of llamafile—like executing across various systems—and the latest functionalities in llama.cpp, including Qwen3.5 models for vision tasks and enhanced tool calling through the Anthropic Messages API. Users can operate these diverse models from a single executable file, streamlining the deployment of AI applications. This update is particularly noteworthy as it reinstates original features alongside delivering improved performance across multiple operating systems and CPU architectures. Key enhancements in version 0.10.0 include support for Metal and CUDA GPU, terminal chat interfaces, and optimizations for various CPU architectures. Additionally, llamafile promises future updates that prioritize user feedback, aim for feature parity with its predecessor, and enhance model bundling capabilities. With engaging options for experimentation across a range of AI models, llamafile 0.10.0 positions itself as a powerful tool for developers and researchers in the AI/ML community.