Google has introduced 'Gemini 3.8 Flash', its latest workhorse AI model focused on coding, agentic workflows, and complex reasoning, just three weeks after launching Gemini 3.7 Flash.
Alongside it, Google has also unveiled 'Gemini 3.8 Flash Cyber', a specialised model designed to help trusted defenders discover and fix software vulnerabilities.
Google describes Gemini 3.8 Flash as its most intelligent workhorse model yet. The company says the new model delivers major improvements over Gemini 3.7 Flash while keeping the same introductory pricing, pushing more advanced reasoning and coding capabilities into a faster, lower-cost model.
Gemini 3.8 Flash: More Reasoning and Better Coding
Gemini 3.8 Flash is designed for long-horizon software engineering and autonomous AI agents. Google says the model can handle complex engineering problems end-to-end and approaches the performance of more expensive frontier models on several tasks.
On the DeepSWE v1.1 long-horizon software engineering benchmark, Gemini 3.8 Flash outperformed most larger frontier models while costing a fraction as much.
It also recorded stronger results than Gemini 3.7 Flash on benchmarks including Vals Finance Agent V2 and Harvey's Legal Agent Benchmark.
Google says the model achieved 54.9% on HLE-Verified, highlighting its ability to work through multi-step problems spanning STEM, humanities and professional fields.
One of the biggest changes is how Gemini 3.8 Flash approaches difficult problems. Rather than simply producing an answer quickly, the model can spend additional compute on complex tasks, take more reasoning steps, and repeatedly call tools when necessary.
This can improve performance, though it may also increase token usage on demanding workloads.
Gemini 3.7 Flash vs Gemini 3.8 Flash
The jump from Gemini 3.7 Flash to 3.8 Flash focuses less on basic speed and more on depth, reliability, and autonomous task completion.
The key difference is that Gemini 3.8 Flash is designed to handle difficult problems. Developers can still use Gemini 3.7 Flash when compute efficiency is the priority, while 3.8 Flash targets workloads where deeper reasoning and better autonomous performance matter more.
Gemini 3.8 Flash Cyber Targets Cybersecurity
Google has also launched Gemini 3.8 Flash Cyber, a specialised cybersecurity model available to trusted defenders through the company's new Fairwind Program.
Unlike a general-purpose coding model, the Cyber version focuses on vulnerability discovery, analysis, and automated patching. Google says it prioritised defensive capabilities over offensive exploitation.
On the CyberGym benchmark, Google says Gemini 3.8 Flash Cyber delivers frontier-level performance in autonomous vulnerability discovery. Its internal testing across complex codebases covering 20 programming languages produced a success rate of more than 70%.
For automated patching, the model achieved a 47.2% pass@1 score on CWE-Bench, close to a leading frontier model's 47.8%, while costing significantly less, according to Google.
Google is already using the model internally. The Chrome Security team found that Gemini 3.8 Flash Cyber produced 2.6 times more correct patches for Chrome vulnerabilities than much larger commercial models.
Google's Cloud Vulnerability Research team also used the model to identify a critical foundational vulnerability in less than two hours, a task the company says would normally take months.
Availability and Pricing
Gemini 3.8 Flash is available to Google AI Pro and Ultra subscribers through the Gemini app, AI Mode in Google Search and Gemini in Google Sheets.
Developers can access it through Google Antigravity, Google AI Studio, the Gemini API and Android Studio, while enterprise customers can use it through Gemini Enterprise.
The introductory API price remains $0.75 per million input tokens and $3.75 per million output tokens. Google says this introductory pricing will remain in place until December 31, 2026, after which the rates will increase to $1.50 per million input tokens and $7.50 per million output tokens.
Gemini 3.8 Flash Cyber, meanwhile, is not being offered as a general consumer model. Access is being provided through the Fairwind Program to trusted government authorities, critical infrastructure operators, software maintainers and other approved defenders.
Google Steps Up AI Competition
The release comes as Google continues to accelerate its AI model development. Gemini 3.8 Flash arrives only three weeks after Gemini 3.7 Flash and represents Google's latest attempt to compete with increasingly capable coding and agent-focused models from other major AI companies.
With stronger long-horizon coding, deeper reasoning and greater autonomy while retaining Flash-level pricing, Gemini 3.8 Flash is positioned as Google's latest push to make frontier-style AI capabilities more accessible to developers and businesses.
At the same time, Gemini 3.8 Flash Cyber gives Google a more specialised offering for organisations looking to use AI for defensive cybersecurity at scale
