|

Google Releases Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber: A Cheaper, More Token-Efficient Flash Tier Built for Agentic Workloads

ByRicardo

Developers constructing manufacturing brokers want larger token effectivity, decrease latency, and extra dependable efficiency. Today, Google has released three new Gemini models. The lineup is Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. All three sit within the Flash tier, which Google tunes for velocity, price, and high-volume agentic work slightly than most reasoning depth.

Gemini 3.6 Flash: higher high quality, fewer tokens, cheaper price

Gemini 3.6 Flash is the brand new default workhorse. It builds on 3.5 Flash and targets coding, data work, and multimodal duties. The most important level is effectivity. On the Artificial Analysis Index, 3.6 Flash makes use of 17% fewer output tokens than 3.5 Flash. On the DeepSWE benchmark by Datacurve, Google experiences as much as a 65% discount. The mannequin additionally takes fewer reasoning steps and software calls per multi-step workflow.

Pricing strikes down alongside effectivity. Gemini 3.6 Flash is priced at $1.50 per 1M enter tokens and $7.50 per 1M output tokens. The output fee drops from the earlier $9.00 on 3.5 Flash. Lower verbosity and a decrease output worth reduces the full price per agentic activity.

Quality positive factors accompany the effectivity positive factors. On DeepSWE, 3.6 Flash scores 49% versus 37% for 3.5 Flash. On MLE Bench, it reaches 63.9% versus 49.7%. On OSWorld-Verified, it hits 83.0% versus 78.4%. On GDPval-AA v2, a knowledge-work benchmark, it scores 1421 versus 1349. Computer use is now a built-in client-side software by means of the Gemini API and Gemini Enterprise. Early prospects together with Hebbia and Harvey cite positive factors in doc parsing, chart and knowledge evaluation, and report drafting.

Google is delivery 3.6 Flash with enhanced Frontier Safety safeguards. These cowl Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber-offense misuse. Full particulars are within the 3.6 Flash model card.

The interactive explainer beneath enables you to examine every mannequin towards its predecessor and estimate token price at your personal quantity.



Gemini 3.5 Flash-Lite: the quickest mannequin within the 3.5 line

Gemini 3.5 Flash-Lite is highlighted for low-latency and high-throughput jobs. Target use circumstances embrace agentic search and doc processing. As measured by Artificial Analysis, it runs at 350 output tokens per second. Pricing is $0.30 per 1M enter tokens and $2.50 per 1M output tokens.

The mannequin clears the prior 3.1 Flash-Lite by extensive margins. On Terminal-Bench 2.1, it scores 54% versus 31%. On GDM-MRCR v2, a long-context benchmark, it reaches 72.2% versus 60.1%. On GDPval-AA v2, it scores 1140 versus 642. Notably, Flash-Lite additionally beats the older 3 Flash on some evals. It leads on SWE-Bench Pro at 54.2% versus 49.6% and on OSWorld-Verified at 74.0% versus 65.1%.

Flash-Lite exposes configurable considering ranges: minimal, low, and larger. Developers can prioritize low-cost, low-latency execution for high-volume duties. They may also interact larger considering ranges for multi-step subagent workloads. Computer use is a built-in software right here too.

Gemini 3.5 Flash Cyber in CodeMender: low-cost brokers that discover and patch bugs

Gemini 3.5 Flash Cyber is essentially the most specialised launch. It is constructed on 3.5 Flash and fine-tuned to search out, validate, and patch software program vulnerabilities. The design premise is the search-space downside. Finding deep flaws means exploring an immense execution search house. A single name to at least one huge mannequin turns into a bottleneck.

The reply is an affordable mannequin referred to as many instances. Inside CodeMender, Google’s code-security agent, a number of 3.5 Flash Cyber brokers run in parallel. CodeMender invokes the mannequin as much as 5 instances, then merges the sub-agent findings into one report. On the CyberFitness center benchmark, this setup reaches aggressive efficiency towards a lot bigger fashions.

The inner evaluations are putting. On Google’s Big Sleep analysis, Flash Cyber considerably surpassed mainline 3.5 Flash and 3.6 Flash. On the V8 JavaScript engine, it discovered 55 distinctive confirmed points at a set variety of invocations. That compares to 47 for mainline 3.5 Flash and 36 for Claude Opus 4.6. It caught 10 points the opposite two fashions missed. In one real-world check, Google’s Cloud Vulnerability Research crew used it to search out remote-code-execution flaws in public APIs inside two hours.

https://weblog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/

Community Reaction

Reaction cut up alongside predictable strains. Builders welcomed the value and effectivity. The delayed flagship drew the loudest criticism. On Hacker News, some argued Google is over-selling capability it can not reliably provision, citing irritating hands-on coding classes. The gated Flash Cyber launch opened a dual-use debate about who ought to maintain automated exploit-finding instruments.

The dashboard beneath aggregates that dialogue by platform. It is a qualitative editorial synthesis, not a scraped dataset, and the tactic be aware is embedded.



Availability

Gemini 3.6 Flash and 3.5 Flash-Lite can be found beginning as we speak. Developers can entry them by means of the Gemini API by way of Google AI Studio and Android Studio. Gemini 3.6 Flash can also be in Google Antigravity and rolling out in GitHub Copilot. Enterprises get each fashions within the Gemini Enterprise Agent Platform, with 3.6 Flash within the Gemini Enterprise app. Everyone can use them by way of the Gemini app, and 3.5 Flash-Lite is rolling out in Google Search. Start with the Developer Guide.

Key Takeaways

  • Gemini 3.6 Flash cuts output tokens by 17% (as much as 65% on DeepSWE) and drops the output worth from $9.00 to $7.50 per 1M.
  • Gemini 3.5 Flash-Lite runs at 350 tokens/sec for $0.30/$2.50 per 1M and beats the older 3 Flash on SWE-Bench Pro and OSWorld-Verified.
  • Gemini 3.5 Flash Cyber powers CodeMender with low-cost multi-agent scans; it discovered 55 distinctive V8 points versus 47 and 36 for 3.5 Flash and Opus 4.6.
  • Flash Cyber is gated to governments and trusted companions beneath a limited-access pilot resulting from dual-use danger.

The publish Google Releases Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber: A Cheaper, More Token-Efficient Flash Tier Built for Agentic Workloads appeared first on MarkTechPost.

Similar Posts