Google Releases Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber: A Cheaper, More High-Performance Flash Tier Built for Agentic Workloads
Developers building production agents need high token efficiency, low latency, and high reliability. Today, Google released three new Gemini models. The list is Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. All three reside in the Flash category, which Google favors for speed, cost, and high-volume agency work rather than deep thinking.
Gemini 3.6 Flash: better quality, fewer tokens, lower price
Gemini 3.6 Flash is the default new workhorse. It builds on Flash 3.5 and targets coding, information processing, and multimodal operations. The main point is efficiency. In the Artificial Analysis Index, 3.6 Flash uses 17% fewer output tokens than 3.5 Flash. In Datacurve’s DeepSWE benchmark, Google reports a reduction of up to 65%. The model also assumes several logical steps and tool calls for multi-step workflows.
The price goes down in line with the efficiency. Gemini 3.6 Flash is priced at $1.50 per 1M input tokens and $7.50 per 1M output tokens. The output rate is down from the previous $9.00 for 3.5 Flash. Low verbosity and low output price reduce the total cost per agent activity.
Quality benefits go hand in hand with efficiency benefits. In DeepSWE, 3.6 Flash scores 49% compared to 37% for 3.5 Flash. In MLE Bench, it reaches 63.9% compared to 49.7%. In OSWorld-Verified, it reaches 83.0% compared to 78.4%. In GDPval-AA v2, a knowledge work benchmark, it scores 1421 compared to 1349. Computing is now a built-in tool for the customer with Gemini API and Gemini Enterprise. Early customers including Hebbia and Harvey cite benefits in document classification, charting and data analysis, and reporting.
Google is shipping 3.6 Flash with improved Frontier Safety protections. This includes Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber-offense. Full details are on the 3.6 Flash card model.
The functional descriptor below allows you to compare each model to its predecessor and measure the cost of tokens by your volume.
‘; html+=’
‘+d.lead+’
‘; html+=’
‘; html+=’
New modelThe predecessorBasic marker
‘; html+=’
‘+d.note+’
‘; panel.innerHTML=html; requestAnimationFrame(function(){panel.querySelectorAll(‘.fill’).forEach(function(f){f.style.width=f.getAttribute(‘data-w’)+’%’;});postSize();}); } var tabs=document.querySelectorAll(‘.tab’); tabs.forEach(function
