Google DeepMind released three models on Tuesday: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. Pricing lands at $1.50/$7.50 per million input/output tokens for 3.6 Flash and $0.30/$2.50 for 3.5 Flash-Lite. Both are live now in Google AI Studio and Android Studio via the Gemini API. No price yet for the cybersecurity-focused 3.5 Flash Cyber.
The pricing context matters. Gemini 3.1 Flash-Lite remains Google's cheapest option at $0.25/$1.50 per million tokens, but Google says it runs 2x slower than the new 3.5 Flash-Lite. Enterprises trading cents for speed get a real tradeoff to calculate. Against the broader market, Xiaomi's MiMo-V2.5 Flash ($0.10/$0.30) and DeepSeek's v4-flash ($0.14/$0.28) undercut the entire Google lineup on raw price. The article includes a 29-model comparison table from Xiaomi, DeepSeek, Alibaba Cloud, OpenAI, Anthropic, xAI, and others, ranging from $0.40 to $60.00 total per million tokens.
The sticker price is only part of the cost story. Google designed these models to consume fewer tokens per task, which compounds savings at scale beyond what the per-token rate shows. The title references a 65% token reduction on long-horizon engineering tasks. That methodology, and the incoming Gemini 3.5 Pro, are the reasons to read the full piece.
[READ ORIGINAL →]