Google Ships 3 Gemini Models, Delays 3.5 Pro Again

Google Expands Gemini Lineup While Its Flagship Stalls

Google has rolled out three new additions to its Gemini artificial intelligence family — Gemini 3.6 Flash, 3.5 Flash-Lite, and a security-focused variant referred to as Flash Cyber — even as the long-awaited Gemini 3.5 Pro remains unreleased 135. The move underscores a strategy of incremental, efficiency-focused updates at a time when the company's most powerful model continues to face delays 47.

What the New Models Offer

The newly launched models are designed around token efficiency, faster response times, and lower operating costs rather than raw capability gains 14. Gemini 3.6 Flash and 3.5 Flash-Lite are pitched as lighter, cheaper options intended to improve latency for AI agents and everyday applications, making them attractive for developers who need speed and cost savings over maximum reasoning power 47. The third release, described by some outlets as Flash Cyber or a cybersecurity-tuned model, is aimed specifically at security use cases, reflecting Google's effort to court enterprise customers with specialized tools rather than a single general-purpose system 235.

Coverage varies slightly on how the trio is labeled — some reports refer to the third model as "3.5 Flash Cyber," while others describe it simply as a cybersecurity-fine-tuned model — but all agree the releases emphasize practical performance over headline-grabbing power 235. The New York Times additionally frames one of the new models as among the company's most capable to date, even as the true flagship remains absent 2.

The Missing Piece: Gemini 3.5 Pro

Despite the flurry of new releases, Gemini 3.5 Pro — the model Google originally teased at its I/O developer conference in May — has still not shipped 53. Reporting citing Bloomberg News indicates the flagship model is running months behind its internal schedule, as Google's engineers work to shore up shortcomings, particularly around coding performance, before it is deemed ready for release 6. That delay has become a recurring theme across coverage, with multiple outlets noting that the absence of 3.5 Pro is becoming increasingly conspicuous the longer Google leans on smaller, faster models to fill the gap 347.

Why the Delay Matters

The holdup carries strategic weight because Gemini 3.5 Pro is meant to be Google's answer to top-tier competitors such as OpenAI and Anthropic, both of which have pushed out increasingly capable frontier models in recent cycles 2. Analysts covering the story suggest that by releasing smaller, optimized models in the interim, Google can keep pace on cost and latency benchmarks — areas increasingly important for AI agents and enterprise deployment — even while its most ambitious model remains a work in progress 4. Still, the prolonged wait raises questions about whether Google's broader AI roadmap is proceeding as planned, and whether the company's public messaging around Gemini has outpaced its actual development timeline 3.

A Pattern of Incremental Releases

Taken together, the coverage paints a picture of a company hedging its bets: rather than waiting for a single, fully polished flagship, Google is iterating publicly with smaller models that address specific needs — speed, affordability, and security — while continuing to refine the more ambitious system behind the scenes 157. This approach lets Google maintain visibility and developer engagement in a fast-moving AI market, even if it means acknowledging, implicitly, that its top-tier model isn't ready 67.

Whether this strategy pays off may depend on how much longer Gemini 3.5 Pro's arrival is delayed and how it performs once it finally launches, especially on coding tasks where Bloomberg's reporting suggests Google has been focused on improvements 6. Until then, the newly released Flash-tier models represent Google's effort to keep its AI product line moving forward and remain competitive on efficiency, even as questions persist about the timeline and execution of its flagship ambitions 34.