Google used last week’s I/O conference to announce the Gemini 3.5 model family and two new flagship tools, Gemini Omni and Gemini Spark. The speaker reports that while these systems were framed for broad multimodal and agentic utility, current, demonstrated capabilities remain narrower than the company’s forward-looking promises. The central tension lies in the shift toward server-side autonomous agents—which trade local control for managed persistence—versus the immediate but more limited status of the models currently available to builders.

This is a standard tech-reporting summary that captures the discrepancy between what Google promises and what is actually shipping to developers today. Skip it if you are already tracking the broad strokes of I/O, as it lacks the deeper technical critique or performance data needed to move beyond a basic product announcement.