Gemini 3.6 Flash Is Now Live: $1.50/$7.50/M, 17% Fewer Output Tokens, Computer Use Built In

The Full Triple Launch — Three New Models

Why the Token Efficiency Story Is Bigger Than the Price Cut

Google cut the output price 17%. The effective cost per completed task fell roughly twice that — and on agentic coding workloads, potentially far more. The reason: Gemini 3.6 Flash is much more token efficient than its predecessor. Google's Logan Kilpatrick said the model spends a lot less tokens to deliver better performance. On multi-step agentic workflows where a model makes many sequential calls, 17% fewer tokens per step compounds across the chain. A ten-step agent workflow with 17% fewer tokens per step costs roughly 83% of the old cost per step — before the price reduction. Combined with the price reduction, organizations running high-volume agentic pipelines on Gemini 3.5 Flash could see effective cost reductions of 30-40% or more by switching to 3.6 Flash on the same tasks.

Gemini 3.5 Flash-Lite — The Cheapest Option Gets Better

Gemini 3.5 Flash-Lite at $0.30/$2.50 per million tokens targets high-throughput, low-latency work — agentic search, document processing, and subagent fan-out. Google's published numbers show a large step over the previous 3.1 Flash-Lite: Terminal-Bench 2.1 at 54% vs 31%, GDM-MRCR v2 long-context at 72.2% vs 60.1%, and GDPval-AA v2 at 1140 vs 642. At $2.50/M output, Flash-Lite is the cheapest option in the Gemini lineup for simple, high-volume tasks — and it now matches or beats the previous generation's default tier on several benchmarks. This is the model for routing, classification, and document extraction at scale.

Gemini 3.5 Flash Cyber — Security-Tuned, No Public Access

Gemini 3.5 Flash Cyber is a restricted security specialist for finding and fixing software vulnerabilities. There is no public API pricing or self-serve access. Google DeepMind published a post about its deployment and security evidence on July 21. It is the first publicly acknowledged security-specialist model in the Gemini family — designed for vulnerability research, penetration testing support, and code security review. Access requires Google partnership. No timeline for public availability has been announced.

Gemini 3.5 Pro — Still Missing

The Flash releases are incremental, and the model that would actually contest the top of the market is still absent while rivals ship. In roughly a week, Grok 4.5, three variants of GPT-5.6, and Kimi K3 all launched. Anthropic's Claude family, led by Fable 5, sits atop the public leaderboards Google's Flash tier now trails. Google now says only that 3.5 Pro is testing with partners and will ship when it is ready. Google also announced it has already begun what it calls "our most ambitious pre-training run yet" for Gemini 4 — suggesting the company may be betting on a generational leap rather than iterating on 3.5 Pro to close the capability gap.

How 3.6 Flash Compares to Claude Sonnet 5 and GPT-5.6 Terra

Gemini 3.6 Flash AA Index score 50 from Artificial Analysis, July 21, 2026. Speed from Google API measurements. Claude Sonnet 5 and GPT-5.6 Terra speeds approximate. Note: some community benchmarks show 3.6 Flash at the same score as 3.5 Flash on Artificial Analysis — Google's own benchmarks show improvements. Test on your own workloads before migrating.

Migration Notes — What Changes From 3.5 Flash

Sources: 9to5Google · OfficeChai · Coursiv · Artificial Analysis · EvoLink · Unite.AI · OpenRouter · Related: Our earlier Gemini 3.6 Flash preview → · Gemini 3.6 Flash vs 3.5 Flash vs Luna → · Gemini 3.5 Pro vs Claude Sonnet 5 →