Hourly ·
Google Launches Gemini 3.6 Flash and Two New Variants as Gemini 4 Training Begins
Google ships three new Gemini models — 3.6 Flash with 17% fewer output tokens, a blistering 3.5 Flash-Lite, and a cybersecurity-focused Flash Cyber — while confirming Gemini 4's most ambitious pre-training run is already underway.
Google dropped a trio of new Gemini models today, headlined by Gemini 3.6 Flash, alongside the ultra-fast 3.5 Flash-Lite and a specialized cybersecurity variant, 3.5 Flash Cyber. The release comes with an unexpected bonus: the company confirmed it has already kicked off its "most ambitious pre-training run yet" for Gemini 4.
Gemini 3.6 Flash is the new workhorse. Google claims it reduces output token usage by 17% compared to 3.5 Flash, and in software engineering benchmarks like DeepSWE, the reduction hits up to 65% — meaning fewer wasted tokens on verbose code edits. Pricing lands at $1.50 per million input tokens and $7.50 per million output tokens, undercutting the previous 3.5 Flash. On knowledge work benchmarks, it scores 1421 on GDPval-AA v2, up from 1349. Computer use is now a built-in client-side tool via the Gemini API.
Gemini 3.5 Flash-Lite is built for raw speed: 350 output tokens per second, making it Google's fastest 3.5-class model. It significantly outperforms prior Flash-Lite generations in agentic workflows — the kind where latency kills the user experience.
Gemini 3.5 Flash Cyber arrives paired with CodeMender, Google's code security agent. The model is purpose-built for cybersecurity applications, tuned for threat detection and code-vulnerability triage. Google positions it as reaching competitive frontier performance when orchestrated through its agent infrastructure.
Conspicuously absent from the lineup is Gemini 3.5 Pro, which Google said is "currently testing with partners" and will ship "as soon as it's ready." The continued Pro delay, now stretching months past the Flash series, has raised questions about whether Google is quietly reorganizing its mid-tier strategy around the forthcoming Gemini 4.
Sources: Google Blog TechCrunch
谷歌推出Gemini 3.6闪存及两款新变种,Gemini 4训练开始
谷歌推出三款新Gemini模型——搭载3.6 Flash的版本比输出tokens减少了17%,性能惊人[K 的3.5 Flash-Lite,以及专注于网络安全的Flash Cyber——同时确认Gemini 4的最雄心[K 勃勃的预训练工作已经开始。
← 小时精选 小时制 · 2026-07-21 20:00 UTC Google发布Gemini 3.6闪充版及两款新[K 变种,Gemini 4训练启动 Google发布了三款新的Gemini模型——3.6 Flash版本减少了1[1D[K 17%的输出令牌、速度惊人的3.5 Flash-Lite以及专注于网络安全的Flash Cyber—同时[K 确认Gemini 4最雄心勃勃的预训练工作已经开始。Google发布了三款新的Gemini模型,[K 其中包括Gemini 3.6闪充版,减少了17%的输出令牌,并且推出了两款新变种——3.5 Fl[2D[K Flash-Lite和专注于网络安全的Flash Cyber。同时,谷歌确认Gemini 4的最雄心勃勃[K 的预训练工作已经开始。
More Hourlies Stories
Content on Anagnorisis is summarized, paraphrased, and editorialized from publicly available sources for length and clarity. Original sources are linked where available. All trademarks belong to their respective owners.
