Google Ships Three New Gemini Models, All Fast and Cheap, and Skips the One Everyone Wanted
Google released Gemini 3.6 Flash, 3.5 Flash-Lite, and a cybersecurity model, aiming for speed and low cost. But the flagship Pro update is still missing.
Google has released three new Gemini AI models at once, all built around the same idea: do more, faster, for less money. The headline model is Gemini 3.6 Flash, which Google calls its “workhorse” (the model meant for everyday, high-volume work). Alongside it come 3.5 Flash-Lite, pitched as the cheapest and fastest option, and 3.5 Flash Cyber, a specialist tuned for security work.
A quick tour of what changed. Gemini 3.6 Flash is better at coding, general knowledge tasks, and handling images and other media, while using up to 17 percent fewer tokens than the model it replaces. Tokens are the small chunks of text a model reads and writes, and since you pay per token, fewer tokens means a lower bill. Flash-Lite is the speed champion, generating around 350 words worth of output per second. Flash Cyber is the unusual one: it is fine-tuned to find and fix software security holes, and Google is keeping it on a tight leash, offering it only to governments and trusted partners through a limited pilot, given how easily such a tool could be misused. The first two models are available now through Google’s developer tools, including the Gemini API and AI Studio.
Here is the part worth noticing. Google shipped cheaper, faster models, but not a new version of Gemini Pro, its top-tier model for the hardest reasoning and coding. Pro has not been updated since February, and Bloomberg reported that the next version has been delayed because it was not hitting Google’s own internal performance targets. Meanwhile rivals have not slowed down: OpenAI has rolled out GPT-5.6 and Anthropic has shipped several Claude models this summer. Google’s product lead Logan Kilpatrick said the company is testing the new Pro with partners and has already begun its “most ambitious” training run yet for Gemini 4. Reading between the lines: Google is keeping its cheap, practical models fresh while the big flagship cooks a little longer.
What this means for you: If you use AI through an app or assistant, you may notice snappier responses and, over time, lower prices, since Flash models power a lot of the tools built on Google’s platform. If you build things with AI, 3.6 Flash is an easy upgrade for coding and agent work where speed and cost matter more than raw brainpower. And if you were holding out for Google’s most capable model to catch up with GPT-5.6 and Claude, the honest answer today is: not yet, keep waiting.
Sources
Google Just Switched Off the 'Creativity Dial' in Its Newest AI, a Good Moment to Learn What It Did
Gemini's newest models now ignore the temperature, top_p, and top_k settings. Here is what those knobs actually did, and why turning them off is part of a bigger trend.
