Google has expanded its Gemini AI portfolio with the launch of Gemini 3.6 Flash, Gemini 3.5 Flash Lite, and Gemini 3.5 Flash Cyber, introducing faster performance, improved coding capabilities, enhanced agentic workflows, and lower operating costs for developers and enterprises.
The company also confirmed that Gemini 3.5 Pro is currently under testing as it continues to build its next generation of AI models.
Gemini 3.6 Flash: Faster, Smarter And More Efficient
Gemini 3.6 Flash is the successor to Gemini 3.5 Flash and has been developed based on feedback from developers and enterprise customers.
According to Google, the model offers notable improvements in coding, knowledge-intensive tasks, multimodal reasoning, and AI agent performance while reducing operational costs.
The company said Gemini 3.6 Flash consumes 17 per cent fewer output tokens than Gemini 3.5 Flash, making it significantly more efficient. It also requires fewer reasoning steps and tool calls to complete complex multi-step workflows, helping developers build AI agents at a lower cost.
Google has priced Gemini 3.6 Flash at $1.50 per one million input tokens and $7.50 per one million output tokens, making it more affordable than its predecessor.
Performance Improvements
Google highlighted several benchmark gains for Gemini 3.6 Flash, including:
Up to a 65 per cent performance improvement on DeepSWE by DataCurve.
49 per cent score on DeepSWE compared to 37 per cent for Gemini 3.5 Flash.
63.9 per cent on MLE Bench versus 49.7 per cent previously.
83 per cent on OSWorld Verified, improving from 78.4 per cent.
Higher knowledge work performance with GDPval-AA v2 score of 1421, compared to 1349 for Gemini 3.5 Flash.
The company also said the model delivers more precise code generation with fewer unnecessary edits and execution loops.
Enhanced Safety Features
Google stated that Gemini 3.6 Flash includes stronger frontier safety protections against misuse involving chemical, biological, radiological, nuclear (CBRN), and cybersecurity threats.
According to the company, the model has become substantially more resistant to jailbreak attempts while reducing unnecessary refusals for legitimate use cases.
Computer Use is now available as a built-in client-side tool through the Gemini API and Gemini Enterprise, enabling AI agents to perform complex tasks more efficiently.
Gemini 3.5 Flash Lite Targets High-Volume AI Workloads
Alongside Gemini 3.6 Flash, Google introduced Gemini 3.5 Flash Lite, describing it as the fastest and most affordable model in the Gemini 3.5 family.
Designed for low-latency applications such as document processing, AI-powered search, and large-scale agentic systems, the model can generate up to 350 output tokens per second, according to Artificial Analysis.
Google has priced Flash Lite at $0.30 per one million input tokens and $2.50 per one million output tokens, offering a strong balance between cost and performance.
Developers can configure the model to prioritise low latency and reduced costs or increase reasoning levels for more demanding multi-agent workloads.
Benchmark Gains For Flash Lite
Google said Gemini 3.5 Flash Lite delivers significant improvements over previous Flash Lite models, including:
54 per cent on Terminal Bench 2.1 compared to 31 per cent.
72.2 per cent on GDM-MRCR v2 versus 60.1 per cent.
1140 on GDPval-AA compared to 642 previously.
54.2 per cent on SWE Bench Pro, outperforming Gemini 3 Flash's 49.6 per cent.
74 per cent on OSWorld Verified compared to 65.1 per cent for Gemini 3 Flash.
The company said these improvements make Flash Lite a stronger option for coding, long-context reasoning, and real-world AI agent execution.
Gemini 3.5 Flash Cyber Focuses On Cybersecurity
Google also introduced Gemini 3.5 Flash Cyber, a specialised AI model built specifically for cybersecurity applications.
The model is based on Gemini 3.5 Flash and has been fine-tuned to identify and fix software vulnerabilities at a lower cost than larger AI models.
Google said Flash Cyber powers its CodeMender security platform, where multiple AI agents collaborate to generate comprehensive vulnerability reports. On the CyberGym benchmark, the model delivers competitive frontier-level performance.
Initially, Gemini 3.5 Flash Cyber will be available only to governments and trusted partners through a limited-access CodeMender pilot programme, allowing security teams to identify and address critical vulnerabilities before they can be exploited.
Availability
Google confirmed that Gemini 3.6 Flash and Gemini 3.5 Flash Lite are available immediately through:
Google AI Studio
Gemini API
Android Studio
Gemini Enterprise
Gemini Enterprise App
Gemini App
The company also announced that Gemini 3.5 Flash Lite is beginning to roll out in Google Search.
Meanwhile, Gemini 3.5 Pro remains in testing and is expected to launch at a later date as Google continues expanding its next-generation AI model lineup.
