Google Launches Gemini 3.8 Flash, Positioning AI Model as Cost-Effective Coding and Agent Workhorse

Google has released Gemini 3.8 Flash and a specialized cybersecurity variant, Gemini 3.8 Flash Cyber, in early September 2026, marketing the model as its most intelligent Flash-class offering yet and demonstrating substantial performance gains over its predecessor. The standard 3.8 Flash model achieves 73.7% on DeepSWE v1.1, a long-horizon software engineering benchmark, approaching the 74.0% score of Claude Opus 5 while maintaining Flash-level pricing and speed. On Terminal-Bench 2.1, the model scores 90.8%, up from 3.7 Flash's 81.6%, and generates output at 311.2 tokens per second—a speed advantage versus many reasoning models in its cost category. The model accepts multimodal input (text, images, audio, video) with a 1-million-token context window and 64-thousand-token output capacity, featuring three adjustable thinking levels to balance reasoning depth and token consumption.