This update introduces a wider selection of AI models and text-to-speech (TTS) options.
You can now use Windows’ built-in speech recognition for text-to-speech transcription without consuming any AI Credits.
We have also switched to a new upstream model provider, giving you access to more affordable models with faster response times. For everyday use, we currently recommend Gemini 3.5 Flash-Lite. When visual capabilities are not required, DeepSeek V4 Flash is another recommended option, offering lower costs and faster responses.