July 22, 2026

AIincider

AI News. No Noise. Just Signal.

Google Ships Gemini 3.6 Flash as Its Pro Model Slips Again

2 min read
Google launched Gemini 3.6 Flash with Computer Use and lower pricing, even as its flagship Gemini 3.5 Pro slips again and Gemini 4 begins. Read more.

Google has released Gemini 3.6 Flash, a faster and cheaper version of its most widely used model, even as the company confirmed that its flagship Gemini 3.5 Pro has slipped yet again. The July 21 launch shows Google leaning hard on its efficient Flash line while its top-tier model keeps missing its targets.

What Google Launched

Gemini 3.6 Flash arrives at $1.50 per million input tokens and $7.50 per million output tokens. Google says it generates about 17 percent fewer output tokens than its predecessor, which lowers the real cost of each request, and it ships with Computer Use built in so the model can click, type, and browse on a user’s behalf.

Google paired it with two more models. Gemini 3.5 Flash-Lite lands at $0.30 input and $2.50 output per million tokens for high-volume, low-cost work. A third variant, Gemini 3.5 Flash Cyber, is a security-tuned model that Google is restricting to governments and trusted partners.

The Missing Flagship

The most striking part of the announcement was what was absent. Gemini 3.5 Pro, the flagship meant to sit at the top of the lineup, has now missed its release window multiple times. Google used the same update to confirm that Pro is still not shipping, after internal testing reportedly found it fell short on coding and complex, long-horizon reasoning.

Rather than dwell on the delay, Google looked past it. The company said it has begun what it calls its most ambitious pretraining run yet, this time for Gemini 4. In effect, it is asking customers to accept a stalled Pro and keep their eyes on the next generation.

Why It Matters

Flash models now do the heavy lifting for most real-world AI applications, where speed and price matter more than a few points on a benchmark. By trimming output tokens and folding in Computer Use, Google is going straight for the agentic workloads that OpenAI and Anthropic are also chasing. At the same time, a repeatedly delayed Pro is a reminder that even the best-funded labs are running into walls at the frontier. The open question is whether Gemini 4 can finally clear the bar that 3.5 Pro could not.

Continue Reading…

Leave a Reply