AI

Google DeepMind launches Gemini 3.8 Flash and 3.8 Flash Cyber for agents and security

Wednesday, September 2, 2026Read Original

Details

  • Google DeepMind unveils two new Gemini models: 3.8 Flash for AI agents and 3.8 Flash Cyber for cybersecurity.
  • Gemini 3.8 Flash is described as DeepMind's most intelligent Flash model to date, with major gains over 3.7 Flash in software engineering, agentic tasks, and multi-step reasoning.
  • Gemini 3.8 Flash Cyber is presented as DeepMind's most capable cybersecurity-focused Gemini model, aimed at securing code and critical systems.
  • The new Fairwind Program gives governments and trusted partners access to 3.8 Flash Cyber to help secure vital infrastructure and protect national security.
  • Gemini 3.8 Flash is rolling out in Antigravity and via the Gemini API through Google AI Studio and the Gemini developer site, extending availability to developers and enterprise users.
  • The announcement positions 3.8 Flash as a core engine for long-horizon AI agents and complex coding workflows, while 3.8 Flash Cyber targets advanced threat detection and secure software development.
  • Governments and critical infrastructure operators are highlighted as early beneficiaries of the Fairwind Program, indicating a focus on aligning the models with national security and public-sector use cases.
  • By building on 3.7 Flash, the 3.8 generation emphasizes improved reasoning, agent orchestration, and hardened code security rather than a completely new architecture.

Impact

The dual launch of Gemini 3.8 Flash and 3.8 Flash Cyber strengthens Google DeepMind’s position in enterprise AI and cybersecurity, directly targeting use cases where coding performance and autonomous agents are becoming central to operations. By formalizing the Fairwind Program for governments and critical infrastructure, Google adds a national-security lens to its AI roadmap, responding to pressure from rivals that are also courting public-sector and regulated industries. The rollout through Antigravity and Google AI Studio lowers friction for developers to adopt these models, likely accelerating agent-based workflows and secure coding practices across Google’s ecosystem.

Rift Dispatch