Google just dropped Gemini 3.8 Flash, its third Flash model release in six weeks. That’s a blistering pace for point releases, but it also tells you something about where Google’s priorities have landed. The company hasn’t shipped a frontier-level Pro model since early 2026, and the promised Gemini 3.5 Pro is looking increasingly like vaporware. Instead of competing at the top of the benchmark tables, Google is flooding the zone with mid-tier models tuned for specific jobs.
This release comes in two flavors. The standard Gemini 3.8 Flash is what Google calls a “workhorse”—aimed at agentic workloads and software development. Then there’s Gemini 3.8 Flash Cyber, built on the same foundation but tuned specifically for vulnerability detection and mitigation. That’s a notable specialization. Security-focused model variants have been a quiet trend this year, and Google shipping one as a standard product suggests enterprise security teams are now a big enough market to justify dedicated model training.
The pricing tells you a lot about the competitive pressure Google is feeling. API access launches at an introductory rate of $0.75 per million input tokens and $3.75 per million output tokens through the end of the year. The regular price will double to $1.50 and $7.50 respectively—but by then, Google will almost certainly have shipped another Flash variant. Other AI labs have been slashing token prices to keep skeptical businesses engaged, and Google is clearly following suit rather than leading.
Benchmark numbers show marginal gains over Gemini 3.7 Flash on most tests, with the biggest improvements in coding evaluations. Google claims the model competes with—or beats—larger, pricier models. That’s the same pitch we heard two weeks ago for 3.7 Flash. The cadence here is starting to feel less like innovation and more like a treadmill. If Google can ship a genuinely better coding model every two weeks, fine—but at some point you have to wonder whether the Pro line is dead or just indefinitely paused.
