They are different shapes of tool. ChatGPT is a destination: you go to it, converse, and carry the result back. OpenCues is a layer over the places you already type: a _ summons the model into your text, including, if you choose, the same models via your existing ChatGPT plan.
What ChatGPT does well
Deep work. Long conversations, reasoning over documents and images, iterating on something substantial. When the task deserves a session, the destination is the right shape, and nothing inline replaces it.
Where the chat window struggles
As a companion to your typing, it is a loop of transfers: switch tab, prompt, send, copy, switch back, paste. The result lands as plain pasted text with no memory of being an AI edit, so partial revert is manual. Nothing surfaces unprompted, and every quick question detours your attention through another interface.
By the numbers
The measured flows: six steps to get an answer into your text and nine to rewrite something you wrote, against one step for either with OpenCues, and the return journey alone (copy, switch back, paste) is three steps against zero. The step counts also measure different ranges: a chat session can do far more than any single fill, which is why the honest framing is complementary, not rivalrous. Counts and sequences in the full comparison.
How OpenCues differs
OpenCues never asks you to leave. The _ fills, transforms, and pulls live data where your cursor already is, edits stay revertable per word, and the openai-subscription provider drives your existing ChatGPT plan, so the two share a bill. Keep the destination for deep sessions; keep a layer for the hundred small moments in between.
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 · Stop copy-pasting from ChatGPT · OpenCues on chatgpt.com · ChatGPT