Google DeepMind dropped three new AI models on Tuesday: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The release is focused squarely on efficiency, speed, and cost, targeting developers building AI agents at scale. But the bigger story is what was not included.
As TechCrunch reported, the update does not include any refresh to Gemini Pro, Google’s flagship model for complex reasoning and coding. That model has not been updated since February, and the AI world has not exactly been standing still in the meantime.
OpenAI has shipped GPT-5.5 and started rolling out GPT-5.6. Anthropic has released Claude Opus 4.8, Claude Sonnet 5, and expanded access to its frontier Fable 5 model. The pace at rival labs has been relentless, which makes Google’s silence on Pro increasingly hard to ignore.
Here is what Tuesday’s release actually includes:
- Gemini 3.6 Flash – Google’s core workhorse model, with better performance in coding, knowledge work, and multimodal tasks. It also reduces token usage by up to 17%, making it cheaper than the previous 3.5 Flash.
- Gemini 3.5 Flash-Lite – The most cost-effective model in the lineup, built for high-volume, budget-sensitive workloads.
- Gemini 3.5 Flash Cyber – A specialized model fine-tuned for finding and fixing cybersecurity vulnerabilities. Access is restricted to governments and trusted partners through a limited pilot program.
The cybersecurity model is an interesting move. Specialized AI tools for security work are a growing area, and Google positioning this behind a controlled access program suggests it is being cautious about how powerful vulnerability-detection tools get distributed. That kind of intentional gating is becoming more common as AI capabilities improve.
The Flash series has always been Google’s answer to the cost-and-speed problem. These models are not meant to compete with the most powerful frontier systems. They are built for production environments where response time and cost per token matter more than raw capability. The 17% token reduction in 3.6 Flash is the kind of improvement that adds up quickly at scale.
The absence of a Pro update is a different matter. Google teased the 3.5 Pro release back in May during the previous Flash launch, saying the model was already being used internally and would roll out the following month. That did not happen. Last week, Bloomberg reported that Google has been struggling with internal delays because the model is not yet meeting its own performance targets.
Google DeepMind product lead Logan Kilpatrick addressed this on Tuesday, saying the team is currently testing Gemini 3.5 Pro with partners and hopes to ship it soon. He also said Google has started its most ambitious pre-training run yet for Gemini 4, which signals the company is already thinking beyond the current generation even as it tries to close the gap on its existing roadmap.
The broader picture here is one of a company balancing two different pressures. On one side, enterprise customers need reliable, cost-efficient models they can build on right now. On the other, the AI race demands visible progress on frontier capabilities. Tuesday’s release addresses the first. The second remains an open question until Pro actually ships.
