Our method: distinguishing facts, citing sources and calmly explaining what can really change for creators, independent and digital professionals.
IA
Google launches Gemini 3.8 Flash and makes it available immediately
Google presents Gemini 3.8 Flash as its new fast general-purpose model for reasoning, code and agentic workflows. The model is available now through the Gemini API and Google AI Studio, as well as in Antigravity, Android Studio, Gemini Enterprise, the Gemini app, AI Mode in Google Search and Gemini in Sheets under eligible plans. The announced introductory price is 0.75 dollar per million input tokens and 3.75 dollars for output, until 31 December 2026. The Gemini 3.8 Flash Cyber variant remains restricted to trusted defenders through the Fairwind programme.
What it can change: Do not migrate a process solely on the promise of benchmarks. Test Gemini 3.8 Flash on a real and repeatable task — research, content production, analysis or automation — then compare quality, correction time and total cost with your current tool.
Read the source: Google →
Marketing
Google signs a multi-year partnership with MrBeast around Gemini
Google and Beast Industries are launching a multi-year collaboration that goes beyond YouTube and combines Gemini and Google Health with MrBeast's content. A first video announced for 5 September will show his team using Gemini in three extreme environments. Google also plans a Gemini campaign with the creator and integration of Fitbit Air into a next challenge. The advertisement does not detail the amount of the contract or its performance indicators.
What it can change: The signal is not to copy Mr Beast's scale, but to design partnerships as a useful demonstration: the product must solve a problem visible in history. In a sponsor folder, propose a usage scenario, an observable result, and several distribution formats.
Read the source: Google →
Marketing
Google recommends to cross attribution, incrementality and media mix modeling
In a new episode Ads Decoded, Google explains that a real-time assignment, one-time incrementality tests and a media mix model can give different readings without any of these tools necessarily being false. The proposed method is to construct a measuring stack combining these three levels, to better capture long shopping routes and to avoid driving only at the immediate ROAS. Google also mentions Meridian, its open source media mix modelling tool, and the role of code assistants to reduce its technical barrier.
What it can change: For a small activity, simply start: keep your source data, follow the delayed conversions and periodically test an isolated campaign or channel. A reliable decision comes from several consistent signals, not from a single allocation table.
Read the source: Google Ads →