Gemini 3.8 Flash raises agent capability at the same token price — but may use more tokens per task

On September 2, 2026, Google released Gemini 3.8 Flash only three weeks after Gemini 3.7 Flash. Google kept the same introductory Gemini API price of $0.75 per million input tokens and $3.75 per million output tokens through December 31, with the published January 1, 2027 rate still doubling to $1.50/$7.50. Google reports materially stronger software-engineering, agentic and multi-step reasoning performance while maintaining Flash-tier speed. The general model is available through the Gemini API, Google AI Studio, Antigravity, Gemini Enterprise and consumer surfaces including AI Mode in Google Search. Google simultaneously launched Gemini 3.8 Flash Cyber, a more permissive cybersecurity variant restricted to approved defenders through the new Fairwind Program.

For production agent systems, 3.8 changes the routing decision again without changing the headline token price. Builders can potentially buy more capability at the same list rate, but per-token pricing is not the same as per-task cost: more aggressive reasoning and longer trajectories can consume more output tokens, and early independent reporting points to that trade-off. The separate Cyber model also shows a new access pattern for frontier dual-use capability: general developers receive the standard model while a more capable security variant sits behind organizational vetting and use restrictions.

Gemini 3.8 Flash launches at the same introductory $0.75/1M input and $3.75/1M output rate as 3.7 Flash. Google says the model materially improves long-horizon software engineering, agentic tasks and specialized reasoning, making the relevant production question total task cost and success rate rather than nominal token price alone.

The Verge reports early analysis indicating 3.8 can use materially more tokens on some tasks even though the per-token rate is unchanged. Teams should therefore compare end-to-end cost, retries, tool calls and wall-clock completion rather than assuming identical list pricing means identical workload economics.

Google says 3.8 Flash is available to developers through the Gemini API and AI Studio, to enterprises through Gemini Enterprise, and to Google AI Pro/Ultra users in the Gemini app, Google Sheets and AI Mode in Search. This makes the release both an API model change and a Search/consumer-model rollout.

Gemini 3.8 Flash Cyber shares the same foundation but has more permissive cybersecurity mitigations. Google limits it to approved Fairwind participants such as governments, critical-infrastructure operators, security partners and software maintainers, with governance requirements including controlled user access and defensive-use restrictions.

The introductory price remains scheduled to expire December 31. Google says standard pricing from January 1, 2027 will be $1.50/1M input and $7.50/1M output, so long-lived agent economics should still be modeled at both current and post-promotion rates.