Google Launches Gemini 3.8 Flash and Flash Cyber: Enhanced Reasoning and Cybersecurity Capabilities

Google launched Gemini 3.8 Flash and the specialized Gemini 3.8 Flash Cyber on September 2, 2026, marking its third Flash model release in six weeks. Positioned as a "workhorse" model, Gemini 3.8 Flash offers introductory pricing of $0.75 per million input tokens and $3.75 per million output tokens, while Gemini 3.8 Flash Cyber is exclusively available to vetted defenders through the Fairwind Program. For broader context, explore our AI News.

Gemini 3.8 Flash: A New Workhorse Model

Gemini 3.8 Flash is designed to be Google's most advanced reasoning and coding model within the Flash series. It follows the release of Gemini 3.7 Flash just three weeks prior, demonstrating Google's rapid development cycle for its Flash line. The model's enhanced performance is attributed to its ability to consume more tokens through additional reasoning steps and iterative tool calls, allowing it to "work harder" on complex tasks.

Initial pricing for Gemini 3.8 Flash is set at $0.75 per million input tokens and $3.75 per million output tokens. However, this introductory rate will increase to $1.50 per million input tokens and $7.50 per million output tokens starting in January 2027. Developers and enterprises can evaluate the model's capabilities and pricing structure before the price adjustment.

Performance Benchmarks and Competitive Landscape

Google has highlighted Gemini 3.8 Flash's competitive performance against other leading models. On the DeepSWE v1.1 benchmark, Gemini 3.8 Flash achieved a score of 73.7%. This places it closely behind Claude Opus 5, which scored 74.0%, and ahead of GPT-5.6 Sol, which recorded 72.7% on the same benchmark. These results suggest that Gemini 3.8 Flash approaches the performance of frontier models at a more accessible cost.

Introducing Gemini 3.8 Flash Cyber for Specialized Defense

Alongside the general-purpose Flash model, Google also unveiled Gemini 3.8 Flash Cyber, a specialized AI designed for cybersecurity applications. This model is not publicly available and is restricted to vetted defenders participating in the Fairwind Program. Its purpose is to assist in identifying and mitigating vulnerabilities across various programming languages.

Internally, Flash Cyber has demonstrated significant success, achieving over 70% in finding bugs across 20 programming languages. Google's Chrome Security team has adopted Flash Cyber internally, reporting that it enables 2.6 times more correct vulnerability patches compared to the best commercial competitor. This specialized model underscores Google's commitment to enhancing digital security through advanced AI.

Key Takeaways for Developers and Security Professionals

• Gemini 3.8 Flash offers improved reasoning and coding capabilities, positioned as a cost-effective "workhorse" model.
• Introductory pricing for Gemini 3.8 Flash is $0.75 per million input and $3.75 per million output tokens, increasing in January 2027.
• The model scored 73.7% on the DeepSWE v1.1 benchmark, comparable to Claude Opus 5 and surpassing GPT-5.6 Sol.
• Gemini 3.8 Flash Cyber is a specialized cybersecurity model available exclusively to vetted defenders via the Fairwind Program.
• Flash Cyber has shown over 70% success in bug finding across 20 programming languages and is used by Google's Chrome Security team for vulnerability patching.

Conclusion

The launch of Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on September 2, 2026, represents Google's continued investment in its AI model ecosystem. Gemini 3.8 Flash aims to provide a powerful and efficient tool for a broad range of AI applications, while Flash Cyber addresses critical needs in the cybersecurity domain. As the introductory pricing for Gemini 3.8 Flash transitions to its regular rate in January 2027, developers and organizations will have the opportunity to integrate these models into their workflows and assess their long-term value. The specialized nature of Flash Cyber also highlights a growing trend of AI models tailored for specific, high-stakes applications.

Sources

• Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
• Proactive cyber defense for governments and enterprises
• Gemini 3.8 Flash Cyber — Google DeepMind
• Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more | The Verge
• Introducing Gemini 3.8 Flash and 3.8 Flash Cyber