Paste long text, clean it into readable sections, summarize each section, and ask questions in-browser with WebGPU. Choose between LFM2.5 230M, 350M, 1.2B-Instruct, and 1.2B-Thinking.