On Tuesday, Google DeepMind released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. Gemini 3.6 Flash is Google’s main workhorse, promising improved capabilities in coding, intelligence, and multimodal productivity, while reducing token usage by up to 17%, making it cheaper than its predecessor, 3.5 Flash.
On Tuesday, Google DeepMind released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. Gemini 3.6 Flash is Google’s main workhorse, promising improved capabilities in coding, intelligence, and multimodal productivity, while reducing token usage by up to 17%, making it cheaper than its predecessor, 3.5 Flash.
Flash 3.6 is a core working model with improved efficiency in coding, analytics, and multimodal tasks. It is cheaper than Flash 3.5 and uses 17% fewer source tokens (and up to 65% in some benchmarks).
3.5 Flash-Lite is the fastest and most cost-effective 3.5 model, generating 350 output tokens per second according to the Artificial Analysis index, and also significantly outperforms previous Flash-Lite generations in agent workflows.
3.5 Flash Cyber is a specialized model designed to find and fix cybersecurity vulnerabilities at a reasonable price. It combines a new, highly efficient, specialized cyber model with the CodeMender code security agent, delivering cutting-edge performance at the cutting edge. According to Google, this model will be available exclusively to governments and trusted partners in a limited pilot program.
In a blog post, the company said that the focus of these releases is on delivering efficiency, low latency, and reliability for customers building large-scale AI agents.
This launch is notable not only for what Google did release — cheaper and faster models optimized for coding, efficiency, and cybersecurity — but also for what it didn’t. The update doesn’t include the long-awaited release of Google’s flagship model, the Gemini Pro, which was last updated in February, TechCrunch reports .
Google announced the Pro release in May during its Flash 3.5 launch, saying that the Pro version «is already in use internally and we look forward to launching it next month.» However, Bloomberg reported last week that Google has faced internal delays in launching 3.5 Pro as the company struggles to meet its performance targets.
Gemini Pro models are typically Google’s most powerful solutions for complex logical thinking and programming tasks, while Flash models are aimed at lower cost and faster response times for work products.
Google DeepMind product manager Logan Kilpatrick said Tuesday that the company is currently testing Gemini 3.5 Pro with partners and hopes to launch it «soon.» He also noted that the team has begun its most ambitious phase of pre-training for Gemini 4.
