Google DeepMind released three new Gemini models on Tuesday, and the most interesting part of the launch is what didn’t ship. According to TechCrunch AI, the company rolled out Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber, but held back the long-awaited update to its flagship Gemini Pro. That’s the model everyone’s been waiting on since February.
What stands out here is the pattern. Google is shipping cheaper, faster, more efficient models while its highest-capability tier sits in delay.
What Google actually shipped
Three models, each aimed at a different job:
- Gemini 3.6 Flash is the new “workhorse.” Google promises better coding, knowledge work, and multimodal performance while cutting token usage by up to 17%, which makes it cheaper to run than the 3.5 Flash it replaces.
- Gemini 3.5 Flash-Lite is the budget option, the most cost-effective model in the class.
- Gemini 3.5 Flash Cyber is a specialist, fine-tuned to find and fix cybersecurity vulnerabilities. TechCrunch AI reports it’ll be limited to governments and trusted partners through a restricted pilot program.
The common thread is efficiency. Google says it’s targeting customers who build AI agents at scale, where latency, cost, and reliability matter more than raw frontier horsepower. Flash models have always been the fast-and-cheap tier; Pro is where the heavy reasoning lives.
The Pro problem
Here’s the part worth paying attention to. Gemini Pro was last updated in February, and Google teased its successor back in May, saying the new Pro was “already being used internally” and would roll out “next month.” That was two months ago.
Last week, Bloomberg reported Google is hitting internal delays on 3.5 Pro because it’s struggling to meet its own performance goals, according to TechCrunch AI. Product lead Logan Kilpatrick said Tuesday the company is testing 3.5 Pro with partners and hopes to “land soon.” He also mentioned the team has kicked off its “most ambitious pre-training run yet” for Gemini 4.
So Google is talking about Gemini 4 before it’s shipped Gemini 3.5 Pro. Read that how you want.
Why this matters
The timing is rough because the competition hasn’t slowed down. Since Google’s last Pro update in February:
- OpenAI shipped GPT-5.5 and started rolling out GPT-5.6.
- Anthropic launched Claude Opus 4.8 and Claude Sonnet 5, and widened access to its frontier Fable 5 model.
That’s the release pace Google is measured against. When your rivals are pushing frontier models every few weeks and your flagship is five months stale, a batch of efficiency-focused Flash models reads as solid engineering but not a headline answer.
The generous take: Google is being disciplined. Shipping a Pro model that misses internal benchmarks would be worse than waiting. Flash models serve most production workloads anyway, and the 17% token savings is real money for anyone running agents at volume. The Cyber model is a genuinely smart niche play, especially with a government and enterprise pilot behind it.
The worried take: a flagship that can’t clear its own bar suggests the frontier is getting harder to push, and Google knows it’s behind on the tier that gets the attention.
What to expect next
For practitioners, a few practical takeaways:
- If you build agents, 3.6 Flash is worth testing now. Cheaper tokens and better coding at the workhorse tier is a direct win for production apps.
- Don’t wait on 3.5 Pro for planning. “Soon” has slipped before. Build against what’s shipped.
- Watch the Cyber pilot. A security-tuned model gated to governments and trusted partners is a signal about where enterprise AI is heading.
The real test comes when 3.5 Pro finally lands and we see whether it closes the gap with GPT-5.6 and Claude Opus 4.8, or just catches up to where rivals were months ago. Until then, Google is competing on price and speed while the frontier race runs without it. More details are available in the original TechCrunch AI report.