Google DeepMind has released two new model variants, Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, which share a foundational architecture but differ in safety envelopes and access controls. The standard Gemini 3.8 Flash is generally available via the Gemini API and other Google platforms, while the Cyber variant is restricted to vetted defenders through the Fairwind Program.
- Gemini 3.8 General Availability: Available on Gemini API, AI Studio, Antigravity, Android Studio, and Gemini Enterprise with closed weights.
- Behavioral Changes: The model executes extra reasoning steps and iterative tool calls for complex tasks, trading higher token consumption for improved accuracy.
- Performance Gains: On DeepSWE v1.1, it outperforms larger frontier models at lower cost, achieving 54.9% on HLE-Verified.
- Cyber Variant Metrics: Flash Cyber achieves 47.2% pass@1 on CWE-Bench and reports 2.6x more correct patches than leading commercial models in Chrome Security tests.
- Pricing: Maintains 3.7 Flash pricing of $0.75/$3.75 per 1M tokens through December 31, 2026.
The release prioritizes vulnerability discovery and patching capabilities for the Cyber variant, offering near-parity with larger models at significantly lower cost while restricting access to trusted organizations.