Serves open-weight LLMs such as Qwen and GPT OSS through an OpenAI-compatible API on wafer-scale hardware, for…