Place in the ecosystem
Gemini in ToTheMoon episodes
In ToTheMoon conversations Gemini shows up as a contender in a model race the hosts refuse to judge by a single benchmark screenshot. Their question is what Google's model actually delivers at work — long context, video analysis, the link with NotebookLM — and whether that is worth switching a subscription. A second thread is personal intelligence: a Gemini tied to Gmail and YouTube history, and a Siri built on Gemini, raise the question of where data is processed and who owns the user's profile.
In late 2025 the hosts covered the week when Gemini 3 reshuffled the leaderboards right after Grok 4.1 had led them, concluding that leaders changing within days make picking “the best model” from a ranking pointless (episode 085). By January 2026 the tone had warmed: the Siri integration was called a big win for Apple, one host admitting, “I like Gemini models more and more” (episode 093), while Gemini's personal intelligence was weighed as a convenience against the sheer scope of access to a person's life (episode 094).
Through 2026 Gemini appears mostly in practical comparisons: in agent pipelines, code writing can be handed to Gemini while another model does the planning (episode 097). In a test of the models' political characters Gemini came out as the most balanced system (episode 129), yet a week later the hosts doubted that the delayed Gemini 3.5 Pro was catching up with OpenAI and Anthropic on complex development (episode 130). Gemini Enterprise for Legal was read as the right wrapper for legal work: a model that can be wrong, next to a base of verified data (episode 157). In autumn Google releases Gemini 4 Argon, and the verdict of episode 182 is restrained: in Ilnar's assessment the model has caught up with Fable, Opus 5.5 and Astra and is only slightly ahead on benchmarks, so dropping everything for Google makes no sense — unlike Gemini 3, which OpenAI answered with a “code red”.