Updated

Light trails looping fast around a small clear glass sphere on a dark background, abstract hero artwork for Gemini 3.6 Flash

Google

Gemini 3.6 Flash

Google's workhorse Flash for production-ready code and fast agent loops, ahead of 3.5 Flash.

Chat with Gemini 3.6 Flash in Super AgentView pricing

What is Gemini 3.6 Flash?

Gemini 3.6 Flash is a Google workhorse model released on July 21, 2026.

Google says it beats Gemini 3.5 Flash at coding, knowledge work and multimodal tasks, writes production-ready code with fewer unwanted edits and debugging loops, and takes fewer reasoning steps and tool calls to finish a multi-step workflow. Computer use is a built-in tool. It reads text, images, video, audio and PDFs with a 1M-token context window.

Provider
Google
Context window
1M tokens
Input
Text, Image, Video, Audio, PDF
Reasoning
Low, medium, high
Max output
65,536 tokens
Knowledge cutoff
March 2026
Released
Jul 21, 2026

How Gemini 3.6 Flash performs

Published by Google

Results Google reported for Gemini 3.6 Flash. Each lab runs its own setup, so compare these with other models only inside the same table.

Benchmark results Google published for Gemini 3.6 Flash
BenchmarkScoreSetting
SWE-Bench Pro (Public)58.7%
DeepSWE v1.149%
Terminal-Bench 2.178%Terminus-2 harness
MLE-Bench63.9%
OSWorld-Verified83%
CharXiv Reasoning85.2%no tools
CharXiv Reasoning89.4%with tools
GDM-MRCR v2, 8 needles91.8%128K, average
GDM-MRCR v2, 8 needles54%1M, pointwise

Source: deepmind.google

What you can do with Gemini 3.6 Flash

  • Production-ready code with fewer unwanted edits, according to Google

  • Fewer reasoning steps and tool calls per workflow than 3.5 Flash

  • Built-in computer use tool

  • Accepts video, audio, images and PDFs as well as text

How to prompt Gemini 3.6 Flash

  • Ask for changes to existing code; Google says it makes fewer unwanted edits.
  • Give it explicit design guidelines when layout matters.
  • Keep effort on low for speed and raise it for multi-step plans.

Where Gemini 3.6 Flash fits best

  • Code changes that should stay small

  • Fast agent loops

  • Questions about clips or audio

  • Long documents up to 1M tokens

How Gemini 3.6 Flash compares

Gemini 3.6 Flash compared with other chat models on Flashloop
ModelLMArena textLMArena codingContext windowReasoning
Gemini 3.6 Flash1483#221518#411M tokensLow, medium, high
Claude Opus 4.81482#251533#181M tokensOff to max
GPT 5.6 Sol1485#211534#171.05M tokensOff to max
GLM-5.31481#261521#331M tokensLow, medium, high, max
GPT-6.1 Sol1480#271537#141.05M tokensLow to max

Good to know before you start

  • Google calls 3.6 Flash a previous-generation model; Gemini 3.8 Flash is its newest Flash
  • Google notes reviewers preferred earlier models for visual layout and styling, so give it design rules
  • In Super Agent, reasoning runs from low to high only

Start chatting with Gemini 3.6 Flash

Chat with Gemini 3.6 Flash in Super Agent

  1. Select Gemini 3.6 Flash

    Open Super Agent on Flashloop and choose Gemini 3.6 Flash in the model picker.

  2. Enter your prompt

    Ask for what you need: an answer, a plan, or the images and videos you want the agent to make.

  3. Send and keep going

    Send the message. Gemini 3.6 Flash answers in Super Agent and runs any generations you asked for.

Questions about Gemini 3.6 Flash

What is Gemini 3.6 Flash designed for?

Coding, knowledge work and agent loops. Google says it writes production-ready code with fewer wasted steps than Gemini 3.5 Flash.

What inputs does Gemini 3.6 Flash accept?

It takes text, images, video, audio and PDFs, up to 1M tokens of context.

Which reasoning levels does Gemini 3.6 Flash support?

Low, medium and high. Super Agent starts it at low.

Gemini 3.6 Flash or Gemini 3.7 Flash?

Gemini 3.7 Flash is the newer of the two. Google says it writes more accurate code on the first pass than 3.6 Flash. Gemini 3.6 Flash is listed under All models.

More chat models on Flashloop

Browse all models