Google just dropped three new Gemini models, but the heavyweight champion everyone was actually waiting for—Gemini 3.5 Pro—is still nowhere to be found. Instead, we got a "Flash" refresh that is quietly changing how we build multi-step AI systems, begging the question: are you overpaying for AI "thinking" when you just need raw scale?

We’ll talk about:

• The 3.6 Flash Workhorse: Why this new middle-weight model is cannibalizing older Pro tiers, boasting a massive 65% token reduction on deep software engineering tasks.
• The "Cheap" Trap: Why blindly picking the budget model (Flash-Lite) can actually cost you more time and money if you don't understand the master-worker agent dynamic.
• The Locked-Down Cyber Model: A look at Gemini 3.5 Flash Cyb ...