Google launches Gemini 3.5 Flash Cyber and updates to 3.6 Flash and 3.5 Flash-Lite
Gemini 3.5 Flash Cyber targets cost-efficient vulnerability detection, while 3.6 Flash and 3.5 Flash-Lite expand coding and multimodal capabilities.
1 source · cross-referenced
- Google introduced Gemini 3.5 Flash Cyber, a security-focused model designed to identify and patch vulnerabilities at lower cost than larger alternatives like Anthropic’s Mythos.
- The model integrates with Google’s CodeMender agent and is initially available to governments and trusted partners.
- Gemini 3.6 Flash and 3.5 Flash-Lite were also released, with improvements in coding, multimodal performance, and cost efficiency.
- Gemini 3.5 Flash Cyber reportedly achieved competitive performance on the CyberGym AI cybersecurity benchmark and identified 55 unique confirmed issues in the V8 JavaScript Engine.
Google announced the release of Gemini 3.5 Flash Cyber, a security-focused model built on Gemini 3.5 Flash and designed to quickly identify and patch software vulnerabilities at lower cost than larger alternatives such as Anthropic’s Mythos.
The model is integrated into Google’s CodeMender, a security-focused coding agent, and will be available first to governments and trusted partners. Google states that CodeMender can invoke 3.5 Flash Cyber multiple times at high speed and low cost, enabling broader code path scanning and vulnerability detection.
In addition to 3.5 Flash Cyber, Google released Gemini 3.6 Flash, which offers improvements in coding and multimodal performance, and 3.5 Flash-Lite, described as the most cost-effective model in the 3.5 series.
Google reports that 3.5 Flash Cyber achieved competitive performance compared to significantly larger models on the CyberGym AI cybersecurity benchmark when invoked up to five times. The model identified 55 unique confirmed issues in the V8 JavaScript Engine, compared to 47 found by Gemini 3.5 Flash and 36 by Opus 4.6, and discovered 10 issues not found by other models.
The company also notes that 3.5 Flash Cyber continued to find new code paths and vulnerabilities after multiple invocations, highlighting its iterative scanning capability.
- Oct 19, 2026 · Hugging Face
Hugging Face adds native PyTorch backend for AWS Trainium via TorchNeuron
Trust79 - Oct 19, 2026 · GitHub · anthropics/claude-code releases
Claude Code v2.1.304 adds guided auto-mode setup, managed agent security checks, and transcript improvements
Trust80 - Oct 17, 2026 · TechCrunch — AI
Vercel reports Jev model reached 35% adoption among paid teams within two weeks of launch
Trust74