Answer
The action is offered only when model details are available. The dialog shows available details about that response, including its model and settings. This is where you inspect the provider used for a completed answer; the composer chooses a type and tier rather than an individual provider. For a limited period, Hashchats temporarily runs most Text AI work on Anthropic Claude models. Ordinary Text answers use Claude Sonnet 5.5, with Gemini 3.8 Flash and then Kimi K3 as backups. Text requests that explicitly ask to think harder or analyze more deeply use Claude Opus 5.5 when no search is needed; if it is unavailable, GPT 6.1 Sol answers, then Gemini 3.8 Flash. Web search answers, translations, native artifacts such as polls, quizzes and charts, and other text-only steps follow the same temporary substitution. Maps-based local answers and reading images, PDFs, video, audio and YouTube links keep their existing providers. Message Details shows the model that actually answered, without a thinking or reasoning suffix on these new responses. Text AI pricing is unchanged. This applies in private AI chats and People chats on web, iOS and Android; there is no separate model or reasoning control to select. Previously saved response labels remain unchanged.