Google Launches New AI Model Gemini 3.8 Flash
Google has announced the Gemini 3.8 Flash model, featuring more reasoning steps and iterative tool calls.
Google has released the new Gemini 3.8 Flash artificial intelligence model, offering advanced reasoning steps for more complex tasks just weeks after its predecessor.
Gemini 3.8 Flash Model Introduced
Google has officially launched its new artificial intelligence model, Gemini 3.8 Flash, shortly after its predecessor, Gemini 3.7 Flash. The company states that this new version works harder on more complex tasks by executing additional reasoning steps and calling tools iteratively.
Cost and Token Usage
The introductory pricing for the model has been set at $0.75 per million input tokens and $3.75 per million output tokens. However, Google emphasizes that the model may use more tokens to maximize performance, which could increase costs.
Industry Experts' Evaluations
Artificial Analysis stated that Gemini 3.8 Flash is the cheapest model measured at this intelligence level. Aigora.ai CEO John Ennis stated that the model delivers high coding quality at a low cost.
Performance and Benchmark Tests
The new model outperformed its predecessor and other frontier models in the DeepSWE v1.1 benchmark test in the field of software engineering. It also managed to surpass its competitors in the Vals Finance Agent V2 and Harvey's Legal Agent tests.
Security Measures and Cyber Program
Google announced that the model comes with security measures to prevent misuse in the areas of Chemical, Biological, Radiological, Nuclear, and cyber attacks. The Gemini 3.8 Flash Cyber model was also announced.
Fairwind Program and Access
Under the Fairwind Program, which is limited to governments and trusted partners, members include CrowdStrike and the Center for Internet Security. Gemini 3.8 Flash is accessible to developers and consumers with a Google AI Pro or Ultra subscription.