Google Ships Its Third Gemini Flash Model in Six Weeks, Still No Sign of Gemini 3.5 Pro
Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, marking its third Flash-series launch in just six weeks, even as the company's promised Gemini 3.5 Pro model remains nowhere in sight.
Google released two new AI models Wednesday, Gemini 3.8 Flash and the specialized cybersecurity model Gemini 3.8 Flash Cyber, adding three Flash series models in six weeks.
Google says 3.8 Flash offers its best reasoning and coding performance yet at the same speed and cost as its predecessor Gemini 3.7 Flash. Gemini 3.8 Flash Cyber is positioned as a frontier-level tool for vulnerability discovery and automated patching.
The releases come as Google relies on its faster and cheaper model line to compete, while its delayed flagship Gemini 3.5 Pro remains absent from the lineup.
Gemini 3.8 Flash Leans on Working Harder, Not Just Faster
Google says Gemini 3.8 Flash’s performance gains come from a deliberate design choice: the model works harder on complex tasks, executing extra reasoning steps and calling tools iteratively rather than optimizing purely for speed.
On DeepSWE v1.1, a benchmark measuring autonomous end-to-end software engineering, Google says 3.8 Flash outperforms most larger, more expensive frontier models at a fraction of the cost.
The model scored 54.9% on HLE-Verified, a test of multi-step reasoning across STEM and professional domains, per the company.
The model is priced identically to its predecessor at $0.75 per million input tokens and $3.75 per million output tokens through the end of the year, as ArsTechnica reports.
It is available immediately through Google Antigravity, AI Studio, and the Gemini app for Pro and Ultra subscribers.
The Cyber Variant Is Already Finding Real Vulnerabilities
Following the AI playbook seen in Anthropic’s Mythos, Gemini 3.8 Flash Cyber limits access to vetted defenders through Google’s Fairwind Program.
Google’s Cloud Vulnerability Research team used it to find a critical foundational vulnerability in under two hours, compared with months of typical manual research, Google says.
Security firm Wiz reported higher recall on its internal penetration testing benchmark while costing several times less than competing frontier models.
Ars Technica linked the launch to Google’s broader credibility push, noting doubts about its standing as a top-tier model maker since Google DeepMind CEO Demis Hassabis, who called for a US-led AI watchdog, moved to a chairman role in early August.
Deputy Koray Kavukcuoglu took over day-to-day leadership under a lower-profile title.
The Real Story Is What Google Still Hasn’t Shipped
The more revealing detail sits outside Wednesday’s announcement.
Gemini 3.5 Pro, which Pichai reportedly told investors would arrive “next month” at Google I/O this May, has now missed three internal target dates.
This leaves Google’s lineup without a high-end reasoning model while Anthropic and OpenAI have shipped multiple frontier releases.
Internal 3.5 Pro candidates were reportedly scrapped for failing to offer a sufficient improvement over existing Flash models, suggesting Google’s mid-tier line may be strong enough to make its premium tier less necessary.
That’s not necessarily bad for users getting capable and cheaper models every few weeks, but it raises questions about Google’s broader strategy.
If Flash continues to absorb improvements that would normally support a Pro release and Gemini 4 remains in post-training, Google may be choosing to compete through speed and price rather than leading on raw capability.
That strategy could work as long as rivals do not release models that Flash cannot match.
