While Google has not released any new state-of-the-art models from the Gemini Pro series, it has demonstrated a strong preference for sequentially releasing various versions of Gemini Flash. Today, Google announced its third Flash model release in just six weeks, an approach that significantly reduces the likelihood of seeing the promised Gemini 3.5 Pro model.
The new Gemini 3.8 Flash model is available in two versions: the standard version, described as a “utility” model for a wide range of agentic tasks and software development, and the “Flash Cyber” version, which shares the same foundations but is optimized for identifying and resolving security vulnerabilities. Google claims that this model is the best-performing model available for reasoning and coding in the Flash tier.
Developer access via the API remains similar to that of the Flash 3.7 version; through the end of this year, access is available at introductory rates of $0.75 per million input tokens and $3.75 per million output tokens. Standard prices will subsequently be $1.50 and $7.50, respectively, although newer models are likely to be released before the price change. This low pricing is most likely a reflection of Google’s strategy to compete with other AI labs that have recently lowered token prices to maintain cautious business operations.