Kerlig and OpenCues agree on the important part: you should bring your own model. Kerlig is a one-time-purchase macOS app that opens a hotkey panel over any app, preloaded with your selected text and hundreds of models including local ones. OpenCues removes the panel: the text field itself is the interface.
What Kerlig does well
A slick, focused Mac experience at a fair one-time price, with very broad provider support. For select-then-act workflows on macOS it is excellent.
Where Kerlig struggles
The flow is select, summon, act, insert: four discrete steps, with the write-back landing as a paste over your selection. That gives it reach into every Mac app at the cost of granularity, edits are wholesale swaps the host app's undo has to unwind, and there is no passive layer watching as you type. macOS only, closed source.
By the numbers
Four steps to a result (hotkey, prompt, submit, Insert) against OpenCues' one, with the Insert click as end-lag; the panel is one new surface over every app, and a pasted result unwinds wholesale. Counts and sequences in the full comparison.
How OpenCues differs
No selection, no hotkey, no panel: type a _ and the fill happens in place, plus transforms, cued alternatives and live data. OpenCues is Apache-2.0 open source where Kerlig is closed, spans terminals, coding agents and Chrome rather than macOS alone, and can also ride your existing Claude or ChatGPT subscription.
Kerlig reaches any native Mac app, which OpenCues integrations do not: they cover terminals, coding agents and Chrome. OpenCues answers with per-word revertable edits and per-feature model routing.
OpenCues is also extensible by design: cues and blanks are markdown files you write yourself, and blanks can bind shell scripts or runtime classes, all under an open standard anyone can implement. See write your own cue and write your own blank.
Related: Full comparison · LLM providers · Kerlig