Google shipped three new Gemini models on Tuesday — 3.6 Flash, 3.5 Flash-Lite, and Flash Cyber — but conspicuously omitted the flagship 3.5 Pro update that developers have been waiting for since February.

Gemini 3.6 Flash serves as Google's "workhorse model" with improved coding and multimodal performance while cutting token usage by up to 17%. The 3.5 Flash-Lite targets cost-conscious applications as the most affordable option in the lineup.

Flash Cyber represents Google's first cybersecurity-specialized model, fine-tuned for vulnerability detection and remediation. The company restricted access to governments and trusted partners through a limited pilot program.

Google positioned the releases around efficiency and reliability for enterprises building AI agents at scale. The focus on operational models rather than frontier capabilities reflects the company's current priorities.

The missing flagship

The absence of Gemini 3.5 Pro becomes more striking when viewed against competitor momentum. OpenAI has shipped GPT-5.5 and begun rolling out GPT-5.6, while Anthropic launched Claude Opus 4.8 and Claude Sonnet 5 in recent months.

Google originally teased the 3.5 Pro release in May, saying it was "already being used internally" with a planned rollout the following month. Bloomberg reported last week that internal delays stemmed from the model failing to meet performance benchmarks.

Gemini Pro models typically represent Google's highest-capability offerings for complex reasoning tasks, while Flash variants prioritize speed and cost efficiency for production deployments.

Google DeepMind product lead Logan Kilpatrick said Tuesday the team is testing 3.5 Pro with partners and hopes to "land soon." He also disclosed that Google has started its "most ambitious pre-training run yet" for Gemini 4.

The delayed flagship release underscores the technical challenges facing Google as it competes with rapidly iterating rivals in the foundation model race.