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

Google has released Gemini 3.8 Flash, its third Flash-tier model in six weeks, Ars Technica reports. The company calls it its best reasoning and coding model yet, even as its promised Gemini 3.5 Pro upgrade remains nowhere in sight.

• The release comes in two versions: a standard "workhorse" Flash model built for agentic tasks and software development, and Gemini 3.8 Flash Cyber, tuned for vulnerability detection and patching
• API access carries an introductory rate through the end of the year, $0.75 per million input tokens and $3.75 per million output, before reverting to a regular $1.50/$7.50
• Gemini 3.8 Flash now tops the DeepSWE leaderboard for software-engineering problem-solving, and Google says it gets there at a lower cost than rivals, though it still trails Claude Opus by a wide margin on the OSWorld-2.0 computer-use benchmark
• On internal testing, Google says Flash Cyber found more vulnerabilities and shipped working patches more often; its Chrome security team reported a 2.6x jump in patch accuracy, and its Cloud team says the model caught a critical vulnerability in two hours

The cadence says as much as the benchmarks. Google has not shipped a frontier-tier Pro model since early 2026, and the company reportedly delayed Gemini 3.5 Pro because its coding performance could not keep pace with competitors. Three Flash releases in six weeks reads like a workaround: ship the cheaper, faster tier constantly, and let its benchmark wins stand in for the bigger release that keeps slipping.

Flash Cyber itself has a short history worth tracking. Intokened covered the 3.5 version of this model when it launched restricted to governments only; the 3.8 release keeps that same limit, with wider access going only to trusted testers and governments, while the standard Flash model rolls out to everyone starting today. The strategy tracks with how Google has run its Flash line all year: a Gemini update cut video-analysis token usage by as much as 88% a few weeks ago, another case of the company chasing efficiency gains on the cheap tier rather than raw capability at the top.

Access still runs through the same channels as every prior Flash release: a Pro or Ultra subscription to reach it inside the Gemini app, or a free account on AI Studio for anyone who wants to test it directly. None of this settles whether Gemini 3.5 Pro ever ships, or whether it still needs to. If Google can keep beating larger, pricier models on coding benchmarks with a Flash-tier release every two to three weeks, the promised flagship starts to look less like a delay and more like a decision.

This article is for informational purposes only and does not constitute investment advice.