Google releases Gemini 3.8 Flash, its third Flash model in six weeks

Google just dropped Gemini 3.8 Flash, its third Flash model release in six weeks. The company hasn't shipped a flagship Gemini Pro model since early 2026, so it's basically betting hard on the Flash line instead—which probably means we shouldn't expect that promised Gemini 3.5 Pro anytime soon. Google's framing this new version as its best reasoning and coding model yet, though the improvements over 3.7 Flash are marginal in most benchmarks, with bigger gains specifically in coding tasks. The model comes in two flavors. Standard Flash handles general work—agents, software development, that kind of thing. Then there's Flash Cyber, built on the same base but tuned specifically for finding and fixing security vulnerabilities. For developers, Google's pricing strategy is aggressive: $0.75 per million input tokens and $3.75 per output million through the end of the year, dropping to $1.50 and $7.50 after that. The introductory rates are basically a necessity at this point—other AI labs have slashed their token prices recently, and Google needs to keep companies from jumping ship as they get more skeptical about AI spending. The benchmark numbers show Gemini 3.8 Flash competing with or beating larger, pricier models, which is the usual pitch from Google. The real story here is the velocity: three releases in six weeks suggests Google is treating Flash as the actual product line while letting Pro languish. That's a shift in strategy, and it signals where the company thinks the market is heading.