Headroom vs Trace

It also checks your driver for the subgroup bug that makes some in-browser LLMs produce garbage.

A macOS menu-bar app that turns any conversation into a clean markdown transcript, with an optional on-device summary. Runs a local speech model entirely on your Mac, in...

About Headroom

It also checks your driver for the subgroup bug that makes some in-browser LLMs produce garbage. One outbound request: ~0.4 MB of public model metadata to derive bytes per token, no run data attached. Projections are measured read bandwidth ÷ bytes touched per token, at empty context; derived manifests make bytes-per-token exact, estimates fall back to an honest band. Real engines land below the ceiling; the gap is engine overhead (see methodology). Runs the transformers.js WebGPU path on the exact model your ceiling is quoted for ( onnx-community/gemma-4-E2B-it-ONNX.

About Trace

A macOS menu-bar app that turns any conversation into a clean markdown transcript, with an optional on-device summary. Runs a local speech model entirely on your Mac, in dozens of languages. The moment is pinned to your transcript at the exact timestamp. Trace provides you with clear markdown transcripts and summaries that you can use anywhere. Either copy the markdown from Trace directly or have it place the file anywhere in your filesystem. Trace can also send files directly to Obsidian, Bear, Drafts or Things with a button press.