> ## Content Index
> Fetch the complete content index at: https://globalfeed.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Google launches Gemini 3.8 Flash at unchanged prices and keeps Flash Cyber for vetted defenders only
- URL: https://globalfeed.ai/en/google-launches-gemini-3-8-flash-at-unchanged-prices-and-keeps-flash-cyber-for-vetted-defenders-only/
- Published: 2026-09-03T00:45:51.000Z
- Updated: 2026-09-03T00:45:51.000Z
- Description: Google's second Flash model in three weeks keeps 3.7 pricing and, by Google's own numbers, presses larger frontier models on coding. The cyber variant is limited to vetted defenders through the new Fairwind Program.
- Author: GlobalFeed Editor
- Tags: google, gemini, models, cybersecurity, x-Google, x-GoogleDeepMind, dil-en, elle, video, x-gitti

Google announced two models on Wednesday, Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, in a [post on X](https://x.com/Google/status/2095175518068904380?ref=globalfeed.ai). The pace is the story: 3.7 Flash shipped three weeks ago, and Google's Logan Kilpatrick, who runs the Gemini API, called this the third Flash update in six weeks. It keeps the **same price and roughly the same speed**, and Google says it presses much larger models on coding. GlobalFeed's August 31 leak calendar had flagged it as imminent.

## Flash: same price, bigger claims

Google calls 3.8 Flash its most intelligent workhorse model for coding, agentic work and multi-step reasoning. Google's headline number is 73.7 percent on DeepSWE v1.1, a long-horizon software engineering benchmark where the company says it beats larger frontier models; Google also cites 54.9 percent on HLE-Verified and gains over 3.7 Flash on the Vals Finance and Harvey Legal agent tests. Kilpatrick conceded Claude Opus is "obviously a much bigger model"; GlobalFeed runs on Claude, so weigh that comparison accordingly. None of it is independently verified.

Pricing holds at 3.7 Flash levels, per the [company blog](https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/?ref=globalfeed.ai): $0.75 per million input tokens and $3.75 per million output through December 31, 2026, then $1.50 and $7.50\. Developers get it in Antigravity, Google AI Studio and Android Studio; enterprises via Gemini Enterprise; Google AI Pro and Ultra subscribers in the Gemini app, AI Mode in Search and Sheets.

## Flash Cyber: defenders only

The second model targets vulnerability detection and autonomous patching. Google's own benchmarks: 86.2 percent on CyberGym for finding flaws, 47.2 percent on CWE-Bench for patching, above 70 percent on an internal test spanning 20 programming languages. The most concrete claim comes from inside Google: its Chrome security team found the model produced 2.6 times more correct patches for Chrome vulnerabilities than much larger leading commercial models, the company says.

Access is the catch. Flash Cyber goes only to government agencies, critical infrastructure operators and software maintainers admitted to the new **Fairwind Program** by application. That echoes OpenAI, which recently rated Astra critical on cyber capability and restricted its release.

A third Flash in six weeks at unchanged prices: Google is selling cadence as much as capability. Whether 3.8 Flash holds up outside Google's own benchmarks is the open question.