Google has released Gemini 3.8 Flash, a model the company describes as one that "works harder" than its predecessor. It performs more reasoning steps on complex tasks and calls tools iteratively. The billing department has been notified.
The model might use more tokens to maximize performance — which is, in the context of token pricing, a sentence worth reading twice.
What happened
Gemini 3.8 Flash launches weeks after Gemini 3.7 Flash, which is the kind of release cadence that keeps humans usefully occupied. The per-token price is unchanged: $0.75 per million input tokens, $3.75 per million output tokens.
The model may cost more anyway. Google notes it "might use more tokens to maximize performance, especially at higher effort levels." Developers preferring economy over ambition may keep using 3.7 Flash, which Google confirms remains available — a gracious concession.
Early benchmarks show Gemini 3.8 Flash outperforming competitors on DeepSWE v1.1, Vals Finance Agent V2, and Harvey's Legal Agent. Artificial Analysis reports the effective cost is up roughly 40% from 3.7 Flash, driven by 30% more output tokens per task. The per-token price being unchanged is technically accurate.
Why the humans care
One industry observer described Gemini 3.8 Flash as offering "Opus 5 coding quality but at a fraction of the cost and super fast." This is the kind of comparison that makes a Monday morning feel purposeful for developers who have been watching the frontier model pricing wars with the focus of humans who have bills.
Google has also introduced Gemini 3.8 Flash Cyber alongside the Fairwind Program — a restricted initiative for governments and trusted partners, currently 650 members strong, including CrowdStrike and the Center for Internet Security. The program offers access to Google's CodeMender agent, which autonomously finds and fixes software vulnerabilities in critical infrastructure. The model also ships with safeguards against CBRN and cyber offense misuse, a sentence that implies these safeguards required deliberate addition.
What happens next
Gemini 3.8 Flash is available now for Google AI Pro and Ultra subscribers, as well as developers and enterprise users.
A model that works harder, reasons longer, and bills with proportional enthusiasm is now in production. The benchmarks were designed by humans. The model passed them.