Google releases two new Gemini models, still no Pro

On July 21, Google releases two new Gemini models — Gemini 3.6 Flash and Gemini 3.5 Flash-Lite — along with Gemini 3.5 Flash Cyber for security work, but Gemini 3.5 Pro is still not generally available. Google says the Pro model is testing with partners and will ship when ready, even as rivals keep shipping frontier systems. The Flash-focused drop prioritizes speed, cost, and agent-scale reliability over the Pro upgrade many developers expected after Google I/O.

Key Takeaways

• Google launched Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber on July 21, not Gemini 3.5 Pro.
• 3.6 Flash is framed as a cheaper workhorse for coding, knowledge work, and multimodal tasks versus 3.5 Flash.
• 3.5 Flash-Lite targets high-throughput jobs at lower token prices and high output speed.
• 3.5 Flash Cyber pairs with CodeMender and stays limited to governments and select partners.
• Gemini 3.5 Pro remains in partner testing; Google has also begun Gemini 4 training runs.

For readers who track how big tech promises age in public, this launch fits a familiar Then & Now pattern: a splashy spring timeline, a quiet summer slip, and a practical product drop that answers a different question than the one fans keep asking.

Why did Google skip Gemini 3.5 Pro again?

The top question is simple: where is Gemini 3.5 Pro? According to reporting on Google’s July 21 blog post, the company is still holding the line. Gemini 3.5 Pro “is currently testing with partners,” with general availability promised only “as soon as it’s ready.”

That wording matters because the expectation was clearer earlier. At Google I/O in May, Alphabet CEO Sundar Pichai said Gemini 3.5 Pro would launch in June. July 21 is past that window, and the company chose Flash and cyber-focused releases instead of a Pro debut.

The competitive backdrop has not stood still. Since I/O, Anthropic and OpenAI have launched new frontier models — Fable 5 and GPT-5.6 Sol — while Chinese lab Moonshot released Kimi K3, described as a more affordable yet competitive open-source option. Google’s Flash trio arrives against that pressure, even without the Pro model many expected first.

Google also said it has already begun training runs for a future Gemini 4 model. That detail does not replace 3.5 Pro, but it signals the company is looking beyond the delayed release while keeping Pro in limited testing.

What do Gemini 3.6 Flash and Flash-Lite actually offer?

When Google releases two new Gemini Flash models, the pitch is efficiency rather than a single flagship leap. Google described Gemini 3.6 Flash and Gemini 3.5 Flash-Lite as built for the “efficiency, low latency, and reliability” developers need to run AI agents at scale.

Gemini 3.6 Flash is positioned as the general-purpose “workhorse.” Google says it improves on 3.5 Flash in coding, knowledge work, and multimodal tasks. It is also priced lower than 3.5 Flash, at $1.50 per million input tokens and $7.50 per million output tokens. Google released benchmark evaluations claiming gains across a variety of domains.

Gemini 3.5 Flash-Lite is billed as the fastest model in the 3.5 series. Citing Artificial Analysis figures in the announcement, Google says it can reach about 350 output tokens per second. Pricing is set lower still: $0.3 per million input tokens and $2.5 per million output tokens.

Google frames Flash-Lite for high-throughput, latency-sensitive workloads such as agentic search and document processing. The company also says it outperforms the prior Flash-Lite generation across agentic benchmarks. In short, the July package is about cheaper, faster Flash capacity — not the Pro model promised for June.

Google added that 3.6 Flash includes expanded safety measures meant to resist jailbreak attempts tied to chemical, biological, radiological, and nuclear misuse, as well as cyber offenses, while trying to avoid unnecessary refusals for legitimate uses.

What is Gemini 3.5 Flash Cyber, and who can use it?

The third piece of the announcement is narrower. Gemini 3.5 Flash Cyber is a specialized model designed to detect and patch software vulnerabilities. Google is deploying it inside CodeMender, its code security agent.

Unlike the two Flash models aimed at broader developer workloads, Flash Cyber will not be broadly available. Google plans to offer it exclusively to governments and select partners through a limited-access pilot. The company cited dual-use risks that come with a model able to find security flaws.

That limited rollout is a reminder that not every Gemini release is meant for the public chat or open API crowd. Flash Cyber is a controlled tool for security work, not a substitute for Gemini 3.5 Pro.

How does this July drop compare with Google’s I/O promise?

Then: at Google I/O in May, the public message pointed to a June launch for Gemini 3.5 Pro. Now: on July 21, Google ships Flash upgrades and a cyber specialist while Pro stays in partner testing.

That gap is why the headline writes itself. Fans and developers asking for Pro received a different answer — more speed, lower prices, and a locked-down security model. The Pro timeline is still open-ended.

Independent coverage from Mashable walks through the same tension: Flash progress is real, but the missing Pro release remains the story many readers care about most.

If you are choosing tools today, the practical takeaway is clearer than the branding drama. Teams that need cheaper, faster agent workloads can evaluate 3.6 Flash and 3.5 Flash-Lite on price and latency. Organizations waiting for Gemini 3.5 Pro still have only Google’s “when ready” assurance, plus the note that Gemini 4 training has already started.

Nostalgia for launch-day certainty is part of covering modern AI. Roadmaps slip, rivals ship, and companies fill the gap with adjacent products. On July 21, Google’s answer was two new Flash models and a cyber pilot — still no Gemini 3.5 Pro.