Google has unveiled a flurry of new Gemini models built to be faster, leaner, and cheaper. But the product push also sharpened the question hanging over its AI strategy: why its promised flagship model still hasn’t arrived.

The company’s latest releases — Gemini 3.6 Flash, 3.5 Flash-Lite, and the security-focused 3.5 Flash Cyber — all point in the same direction: lower cost and higher efficiency for businesses deploying AI at scale. Several outlets framed the move as Google leaning into a market that now cares less about headline benchmarks and more about usable performance per dollar.¹ Axios described the shift bluntly: “The AI deployment race has shifted from benchmark bragging rights to who can provide the best model at the lowest price.”²

Google’s own messaging matches that emphasis. In announcing Flash Cyber, DeepMind called it a “cost-efficient and highly capable alternative to large, costly cybersecurity models,” while limiting early access to governments and trusted partners because of the technology’s dual-use risks.³ Reporting from The Verge and Ars Technica also highlighted the practical pitch: 3.6 Flash uses fewer tokens, Flash-Lite is built for high-volume work, and Flash Cyber is aimed at finding and patching vulnerabilities quickly.⁴⁵

Still, the missing piece is Gemini 3.5 Pro. Google had previously signaled a June release, but it now says the model remains in testing and will be released when ready.⁶ That delay has fueled criticism that Google is winning on efficiency while ceding the “frontier” conversation to rivals. Business Insider noted that competitors have openly mocked the holdup, and one analyst said the delay has shifted perception “from leading edge to trailing edge.”⁷

On X, Google executives tried to keep attention on the new launches. Sundar Pichai amplified a post saying the releases were about “better performance, lower latency, and a smaller bill,” while teasing that “Next up: Gemini 3.5 Pro.”Sundar Pichai (sundarpichai)⁸ Demis Hassabis shared the same message.Demis Hassabis (demishassabis)⁹

What emerges is a split-screen picture: Google may have found a strong commercial argument for cheaper AI, but until 3.5 Pro ships, questions about its place at the cutting edge are unlikely to fade.