Comment by deep1283

The token efficiency improvement might be underrated. If the model solves tasks with fewer tokens, that directly translates into lower cost and faster responses for anyone building on the API.