AIToday
Large Language ModelsAI Coding AssistantsAI Safety & AlignmentGoogle DeepMindPublished: Sep 3, 2026, 04:01 JST

Gemini 3.8 Flash and Cyber models launched

Send on LINE
Gemini 3.8 Flash and Cyber models launched

3 Key Points

  1. What happened

    Google DeepMind launched Gemini 3.8 Flash, a new AI model for coding and reasoning, and Gemini 3.8 Flash Cyber for cybersecurity. Both are available today, with Flash priced at $0.75 per million input tokens and $3.75 per million output tokens.

  2. Why it matters

    This is Google's third Flash release in six weeks, building on 3.7 Flash. The Cyber version shows frontier-level performance in finding vulnerabilities and patching them, often at lower cost and faster speed than larger models, which could help defenders secure code more effectively.

  3. What to watch

    Gemini 3.8 Flash Cyber is only available to trusted defenders through the new Fairwind Program, including government authorities and critical infrastructure operators. For example, it produced 2.6 times more correct patches to Chrome vulnerabilities and found a critical vulnerability in under 2 hours.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Google DeepMind's launch of Gemini 3.8 Flash and Flash Cyber marks the third Flash release in six weeks, indicating a rapid iteration cycle. The core model focuses on long-horizon coding and agentic tasks, showing gains that often approach higher-cost frontier models. The Cyber variant is tailored for defensive security, with emphasis on vulnerability discovery and automated patching over offensive capabilities, aligning with their Frontier Safety Framework's focus on beneficial use cases.

Real-world applications already show promise: the Chrome Security team found 3.8 Flash Cyber produced 2.6 times more correct patches than larger commercial models, and Google's Cloud Vulnerability Research team found a critical vulnerability in less than 2 hours, a task that usually takes months. These results suggest a tangible advantage for defenders if the model's access is scaled appropriately.

The Cyber model's restricted availability through the Fairwind Program highlights the balance between enabling powerful defensive tools and mitigating misuse. By limiting access to trusted entities, Google aims to prevent the model from being used for offensive purposes, a concern given its advanced capabilities in vulnerability discovery.

FAQ
How much does Gemini 3.8 Flash cost?
Gemini 3.8 Flash is priced at $0.75 per million input tokens and $3.75 per million output tokens, the same as 3.7 Flash.
Who can access Gemini 3.8 Flash Cyber?
Gemini 3.8 Flash Cyber is available only to trusted defenders through the Fairwind Program, including government authorities, critical infrastructure operators, and software maintainers.
What was Gemini 3.8 Flash Cyber's performance on patching?
On the CWE-Bench benchmark, it achieved a pass@1 of 47.2% compared to a leading frontier model at 47.8%, but at a significantly lower cost.
Google DeepMindRead Original Article
Send on LINE

Also reported by Ars Technica AI, ITmedia AI+, THE DECODER

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleZeroDrift launches Guard for Agents compliance service