Google has released Gemini 3.8 Flash, a new general purpose AI model aimed at software development, autonomous agents, and complex reasoning tasks. The company is also introducing Gemini 3.8 Flash Cyber, a specialized version for vulnerability detection and code patching that will initially be limited to approved defensive security organizations.
Gemini 3.8 Flash replaces the recently released 3.7 Flash as Google’s flagship low cost workhorse model. Google says the new version offers stronger results in long running coding tasks, professional analysis, and multi step problem solving while retaining the same introductory API pricing: $0.75 per million input tokens and $3.75 per million output tokens. That rate runs through December 31, 2026. From January 1, 2027, prices will double to $1.50 for input and $7.50 for output per million tokens.
Gemini 3.8 Flash is designed to spend more time on difficult requests. It can take additional reasoning steps and repeatedly use available tools before returning an answer. This may improve results in demanding workflows, but it can also increase token use at higher effort settings.
Google advises developers focused mainly on efficiency to choose lower effort settings or continue using Gemini 3.7 Flash.
Built for longer AI workflows
Google positions the model for tasks that require an AI system to plan, act, check results, and continue working over several steps. These so called agentic workflows are increasingly used to write and test software, research information, create reports, and operate within business tools.
On DeepSWE v1.1, a benchmark for complex software engineering work, Google says Gemini 3.8 Flash outperforms most larger frontier models in completing engineering problems independently. The company also reports gains on Vals Finance Agent V2 and Harvey’s Legal Agent Benchmark, which test professional analysis and reporting. It says the model achieves 54.9 percent on HLE Verified, a benchmark covering reasoning in areas including science, humanities, and professional work. Those results are vendor reported benchmark claims and do not necessarily predict performance in every organization’s daily work.
Gemini 3.8 Flash is available through the Gemini API, Google AI Studio, Android Studio, and Google Antigravity. Google also makes it available to Google AI Pro and Ultra subscribers in the Gemini app, AI Mode in Search, and Gemini in Google Sheets. Enterprise customers can access it through Gemini Enterprise.
A separate model for defenders
Gemini 3.8 Flash Cyber targets cybersecurity teams that identify weaknesses in software and create fixes. Google says it trained the shared model foundation with demanding cybersecurity work, then applied a more permissive set of cyber safeguards to the specialized version.
Access is therefore restricted through Google’s new Fairwind Program. The company says trusted government bodies, critical infrastructure operators, and software maintainers can apply. Google emphasizes that it prioritized defensive functions, particularly finding and fixing vulnerabilities, rather than capabilities for exploiting them.
The company says both releases include protections against misuse involving cyber offense and chemical, biological, radiological, and nuclear risks. It also reports improved resistance to prompt injection attacks, citing measurements by security firm Gray Swan.
Sources
- Introducing Gemini 3.8 Flash and 3.8 Flash Cyber – Google Blog
- Google launches Gemini 3.8 Flash, three weeks after 3.7 Flash launch, for an introductory price of $0.75/1M input and $3.75/1M output tokens until December 31 – 9to5Google
Stay up to date
AI for content creation: the latest tools, tips and trends. Every two weeks in your inbox:
