Google releases Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber models with delayed 3.5 Pro

Decision brief

What changed

Google released three new Gemini models—3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber—focused on efficiency, cost reduction, and cybersecurity tasks, while the flagship Gemini 3.5 Pro model, originally promised for June, remains delayed due to internal performance challenges.

Why it matters

The new Flash-tier models offer meaningful cost savings (17% fewer tokens, output pricing down from $9 to $7.50/1M tokens) and speed gains (350 tokens/sec for Flash-Lite), which matters for enterprises scaling agentic workflows on tight margins. However, the Pro delay signals Google may be lagging OpenAI and Anthropic on flagship capability, forcing developers to choose between using mid-tier models now or waiting, which affects vendor selection and roadmap planning for AI-dependent products.

Affected roles

CTO CISO CFO COO

Evidence

Multiple independent outlets (TechCrunch, The Verge, Ars Technica, The New Stack, MarkTechPost) and Google's own DeepMind blog consistently report the same three model names, pricing, and the Pro delay, with TechCrunch and Ars Technica explicitly noting competitive lag versus OpenAI and Anthropic.

What remains uncertain

The specific 'internal performance challenges' causing the Pro delay are not detailed, and no new release date is confirmed; coding benchmark comparisons (49% vs competitors' 54-70%) come from only one outlet (The New Stack) and aren't independently corroborated across all sources. It's also unclear how the Flash Cyber model's real-world vulnerability-detection performance compares to Anthropic's Mythos beyond Google's own framing.

Monitor next

Watch for the actual release date and benchmark performance of Gemini 3.5 Pro, which will indicate whether Google can close the capability gap with OpenAI and Anthropic's flagship models.

Analytical support, not advice — assumptions and open questions stated above.

Source coverage