Google Launches Gemini 3.8 Flash and 3.8 Flash Cyber, Focusing on Agentic Workflows and Cyber Defense
On September 2, 2026, Google announced Gemini 3.8 Flash and the security-focused 3.8 Flash Cyber. This article summarizes their positioning, introductory pricing and access as described in Google's official release post.
News date: About 4 min read
Contents
What Happened
On September 2, 2026, Google announced Gemini 3.8, which includes two versions: Gemini 3.8 Flash and Gemini 3.8 Flash Cyber. Google says it follows 3.7 Flash, released about three weeks earlier, and is the third Flash release in six weeks.
Google calls Gemini 3.8 Flash its smartest workhorse model, with significant improvements over 3.7 Flash in software engineering, agentic tasks (tasks where the AI completes work on its own in multiple steps) and multi-step reasoning in professional domains. Google also says both models are powered by the same underlying intelligence, and that the coding and reasoning gains come from several innovations, including rigorous training in the security domain.
How the Three Models Differ
The table below, based on Google's announcement, summarizes the positioning of the two new models and the still-supported 3.7 Flash. All descriptions are Google's own. A token in the table is the unit a model uses to measure the text it processes; fees are calculated from the number of input and output tokens.
| Model | Google's positioning | Availability or price |
|---|---|---|
| Gemini 3.8 Flash | Smartest workhorse model, emphasizing software engineering, agentic tasks and professional-domain reasoning | Introductory price: $0.75 per million input tokens, $3.75 per million output tokens |
| Gemini 3.8 Flash Cyber | Its most capable security model, emphasizing vulnerability detection and automated patching | Offered to trusted defenders through the new Fairwind Program |
| Gemini 3.7 Flash | Suited to efficiency-first workloads, still fully supported | 3.8 Flash's introductory price is the same as this model's |
Google's Performance Claims
Google cited several benchmarks, which are standardized tests used to compare model capabilities. Google says that on DeepSWE v1.1 (long-horizon software engineering), 3.8 Flash outperforms most larger frontier models (the group of models with the most advanced capabilities today) at autonomously solving complex engineering problems, at a fraction of the cost.
Google also says 3.8 Flash outperforms 3.7 Flash and other frontier models on benchmarks such as Vals Finance Agent V2 and Harvey's Legal Agent Benchmark, and scores 54.9% on HLE-Verified.
Google explains that on complex tasks, 3.8 Flash performs extra reasoning steps and calls tools repeatedly; to achieve the best performance, especially at higher effort levels (a setting for how much reasoning the model invests in a task), the model sometimes uses more tokens.
What It Means for General Readers
Google says that if compute efficiency is the main constraint, developers can use a lower effort level to reduce token overhead, or keep using Gemini 3.7 Flash, which remains fully supported and suits efficiency-first workloads.
On security, Google says 3.8 Flash Cyber delivers frontier-level performance in vulnerability detection and automated patching, and is offered to "trusted defenders" through the new Fairwind Program.
Frequently asked questions
How much does Gemini 3.8 Flash cost?
Google says its introductory price is the same as 3.7 Flash: $0.75 per million input tokens and $3.75 per million output tokens.
What is Gemini 3.8 Flash Cyber, and who can use it?
Google calls it its most capable security model, with frontier-level performance in vulnerability detection and automated patching, offered to trusted defenders through the new Fairwind Program.
Will 3.8 Flash use more tokens?
Google says that to achieve the best performance, especially at higher effort levels, the model sometimes uses more tokens. Developers can switch to a lower effort level to reduce token overhead.
Can I still use 3.7 Flash?
Google says 3.7 Flash remains fully supported and suits efficiency-first workloads.
Are these performance figures reliable?
This article attributes the performance figures to Google's official release post and presents them as the company's claims.
Browse the latest news in this topic
Found this guide helpful? Share it with someone who needs it
Gemini 3.8 Flash Launched: What Model Upgrades Mean for Everyday Tools
Google launched Gemini 3.8 Flash and restricted-access Flash Cyber on September 2, 2026, highlighting coding, agents, and multi-step reasoning. Exploring search, spreadsheets, and data organization, this article distinguishes model upgrades from product entry points, subscription access, developer promo pricing and deadlines, and why identical unit rates might still lead to differing total usage costs.
Signal Blocks Screenshots by Default on Android and iOS 27 to Stop AI Agents Like Siri and Gemini From Reading Your Screen
Claude Code 2.1.329: Code Mode Works With Telemetry, /tasks Lists Scheduled Tasks, macOS Computer Use Gets Stricter Safety Rules
Kimi Code CLI v2.7.0 Adds a Native Amazon Bedrock Provider: Use Your AWS Credentials Instead of a Separate API Key
Sources
Compiled from the references and verification dates listed above; fares, schedules, product features, and terms can change, so please confirm the latest details on the relevant official website before travelling, booking, or purchasing.