Google has launched Gemini 3.8 Flash, arriving only weeks after its predecessor. The company claims the new model “works harder” than Gemini 3.7 Flash by performing more reasoning steps on complex tasks and calling tools iteratively.
The Token and Pricing Dilemma
While the introductory pricing remains identical to 3.7 Flash — $0.75 per million input tokens and $3.75 per million output tokens — the total bill for users could rise. Google warned that “the model might use more tokens to maximize performance,” particularly at higher effort levels. Early measurements from Artificial Analysis suggest that Gemini 3.8 Flash is effectively ~40% more expensive than its predecessor, driven by a 30% increase in output tokens per task and more turns on agentic evaluations.
Benchmarks and Engineering Capabilities
The model offers “significant improvements” for software engineering and autonomous AI agents. It outperformed frontier models, including Anthropic’s Fable 5, on the DeepSWE v1.1 software engineering benchmark. John Ennis, CEO of Aigora.ai, compared the model’s capabilities to “Opus 5 coding quality” but at a fraction of the cost, noting its potential for tasks like creating remotion videos.
Cybersecurity and the Fairwind Program
Alongside the standard release, Google introduced Gemini 3.8 Flash Cyber as part of its new Fairwind Program. Limited to governments and “trusted partners” such as CrowdStrike and the Center for Internet Security, the program provides access to the CodeMender agent. This tool is designed to autonomously find and fix vulnerabilities to protect critical infrastructure and national security. The model also ships with safeguards against misuse in Chemical, Biological, Radiological, and Nuclear (CBRN) domains.
