\n<\/p>\n
A hidden menu in Google App v17.18.22 reveals that Google is testing new AI models for Gemini Live, including a new “Thinking” variant.<\/span><\/p>\n PAUL MONCKTON<\/small><\/div>\n<\/div>\n<\/figure>\n Google has been testing multiple unknown AI models within Gemini Live, hinting at a possible upgrade to the voice-controlled chatbot in time for Google I\/O 2026 later this month.<\/p>\n While investigating an unrelated unreleased feature in the Google app code, a previously unseen settings cog button caught my eye. Enabling the button in the code revealed a new model selector menu that allows the user to change the AI model Gemini Live uses to power its voice chats.<\/p>\n A hidden model selector in the Google App offers a menu seven AI models, including personalization and thinking variants.<\/span><\/p>\n PAUL MONCKTON<\/small><\/div>\n<\/div>\n<\/figure>\n The menu, currently hidden behind a server-side flag, reveals seven AI model options I haven\u2019t seen before, including codenames \u201cCapybara,\u201d \u201cNitrogen\u201d and a specialized personalization model. The displayed menu options are as follows:<\/p>\n Of these, \u201cA2A_Rev25_RC2\u201d and \u201cA2A_Rev25_RC2_Thinking\u201d appeared overnight on May 8, showing that Google now has two new audio-to-audio models at the Release Candidate 2 stage, nearing production readiness. The presence of a Thinking model is particularly interesting, as it suggests a variant with enhanced reasoning capabilities may soon be available.<\/p>\n Currently, Gemini Live uses only one model \u2014 Gemini 3.1 Flash Live<\/u><\/a>, a native input model designed to process raw audio and video streams directly. The existence of multiple new models strongly suggests that Google is trying out some alternatives internally before a public release.<\/p>\n No public information exists for these models, but there are a few small clues in the naming. Here, \u201cA2A\u201d most likely stands for Audio-to-Audio<\/u><\/a>, Google\u2019s term for models that process speech and audio directly, rather than converting them to text first.<\/p>\n In the screenshot above we see a model labeled \u201cP13n<\/u><\/a>,\u201d a shortened form of the word \u201cpersonalization.\u201d This hints at a specialized Gemini variant with additional personalization and behavioral features baked directly into the model.<\/p>\n While the regular Gemini interface currently offers users a choice between Fast, Thinking and Pro models, Gemini Live currently offers no such option.<\/p>\n Switchable models would allow the company to provide a more powerful voice assistant to customers willing to pay for it, or perhaps allow users to trade Gemini Live\u2019s snappy responses for more thoughtful ones that take a little longer.<\/p>\n Neither Capybara nor Nitrogen appear in any prior Google documentation. However, terms like \u201cRev25\u201d and \u201cExp\u201d suggest that the company has already been through several revisions of these models and likely has both stable and experimental versions of the Capybara model under test.<\/p>\n The list of available models is delivered by Google\u2019s servers, meaning the company can add or remove models without an app update.<\/p>\n We don\u2019t know at this point whether Google is about to bring model selection to Gemini Live users, or whether this is purely an internal testing tool. The model selector interface, as it stands, remains unpolished and isn\u2019t ready for release.<\/p>\n My testing confirms that the selected model name is transmitted to Google\u2019s servers when a voice session begins, but it remains unclear whether functionally different models are actually served in response.<\/p>\nKey Takeaways<\/h2>\n
\n
\n
What The Code Reveals<\/h2>\n
The P13n Variant<\/h2>\n
Why This Matters<\/h2>\n
What We Don\u2019t Know<\/h2>\n
Nitrogen And Capybara Variants<\/h2>\n