TL;DR

Google released Gemini 3.6 Flash and Gemini 3.5 Flash-Lite on July 21, 2026. Both models are generally available rather than experimental previews.

Gemini 3.6 Flash is the stronger workhorse for coding, multimodal analysis, knowledge work, and complex agentic workflows.

Gemini 3.5 Flash-Lite is the faster, lower-cost execution model for document extraction, translation, classification, search, and high-volume subagent tasks.

Both models support a 1M-token context window, up to 64K output tokens, multimodal input, thinking, function calling, structured output, and search grounding.

Gemini 3.6 Flash is not dramatically more intelligent than Gemini 3.5 Flash on every benchmark. Its biggest advantages are fewer output tokens, fewer tool calls, shorter task times, and a lower output price.

The most interesting way to use them may not be choosing one model. It may be using Gemini 3.6 Flash as the main agent and Gemini 3.5 Flash-Lite as the execution layer.

Both models are now available through GPTProto at 40% off Google’s list pricing. You can call the Gemini 3.6 Flash API or Gemini 3.5 Flash-Lite API with one GPTProto API key and an OpenAI-compatible endpoint.

Google has released two new production-ready Flash models, but their names do not immediately explain how they differ.

Is Gemini 3.5 Flash-Lite simply a smaller Gemini 3.6 Flash? Does “GA” indicate a separate model? Is Gemini 3.6 Flash actually better than Gemini 3.5 Flash, or is it only cheaper?

The short answer is that Google has created two models for two different layers of an AI system. Gemini 3.6 Flash is designed to make harder decisions and coordinate complex workflows. Gemini 3.5 Flash-Lite is designed to execute large numbers of smaller tasks quickly and economically.

This guide covers everything you need to know about Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, including their release dates, pricing, specifications, use cases, migration requirements, and how Gemini 3.6 Flash compares with Gemini 3.5 Flash and Kimi K3.