Google Quietly Ships Three New Gemini Models While 3.5 Pro Remains MIA

The Three New Models

Gemini 3.6 Flash: The Powerful Workhorse

• Key specs: Google's most capable Flash model to date, described as a "workhorse model" . It delivers improved coding, knowledge work, and multimodal performance, with the key advantage of a 17% reduction in token usage compared to its predecessor, Gemini 3.5 Flash — meaning it is cheaper to run . Official documentation notes it "balances speed with intelligence to deliver strong performance in agentic and multimodal tasks" .
• Target use cases: General-purpose production workloads that need a balance of speed and strong intelligence — especially agentic applications, code generation, knowledge work, and multimodal tasks .

Gemini 3.5 Flash-Lite: The Cost-Efficiency Champion

• Key specs: An updated version of Google's "most cost-efficient lightweight offering" . It is designed for high-throughput execution with minimal latency .
• Target use cases: High-volume, lower-complexity tasks where budget matters more than frontier-level performance — such as classification, basic summarization, or simple data extraction .

Gemini 3.5 Flash Cyber: The Security Specialist

• Key specs: A fine-tuned variant of Gemini 3.5 Flash, purpose-built for cybersecurity workflows . According to the New York Times, it can "find and patch security vulnerabilities at a lower cost than other larger models" . It will be made available exclusively to governments and trusted partners .
• Target use cases: Threat detection, vulnerability analysis, security operations — any domain where a specialized, low-cost model adds value over a general-purpose frontier model .

The Elephant in the Room: Why Gemini 3.5 Pro Is Still Missing

These three launch models sit at the lightweight end of Google's Gemini lineup. The flagship model, Gemini 3.5 Pro, which was promised to arrive in June 2026 by CEO Sundar Pichai at Google I/O, remains unreleased . As of July 22, 2026, Google's official model catalog lists no Gemini 3.5 Pro — the Pro tier is still at Gemini 3.1 Pro (Preview) and Gemini 2.5 Pro (Stable) .

• Coding performance is the bottleneck: A July 16 Bloomberg report, citing 10 current and former employees, revealed that the model's coding performance "fell short of internal goals" . Engineers found it struggled with complex code generation and long-horizon reasoning tasks . In late June, Google updated its training data in an attempt to fix this, but the results were disappointing . DeepMind even scrapped a near-ready version and began rebuilding the model from scratch .
• Multiple missed deadlines: Google initially targeted June 2026. A second target, July 17, passed with no release . A third deadline has come and gone . The has now missed at least three public deadlines .
• Market fallout: The delay wiped roughly $200 billion from Alphabet's market capitalization — a drop of about 4.4% — following the Bloomberg report .

What This Tells Us About Google's AI Strategy

The July 21 release offers a clear strategic picture: Google is shipping what it can ship. All three models are lighter, cheaper, and specialized — Flash variants that can be incrementally improved and launched without the pressure of a frontier-level promise. Meanwhile, its flagship has become a cautionary tale about the difficulty of building truly advanced coding and reasoning models .

The fact that Google shared no new timeline for Gemini 3.5 Pro alongside Tuesday's announcement suggests the backend problems are not yet resolved. The company appears to be buying time by adding value elsewhere — with a cheaper workhorse, an ultra-low-cost variant, and a niche security specialist — while its most important competitive entry languishes.