Google has made Gemini 3.8 Flash generally available for developers, adding a faster model option for coding, multi-step reasoning and software agents. The launch was highlighted by Geeky Gadgets, while Google’s model documentation and a separate Ars Technica report provide the product and availability details.

Google describes Gemini 3.8 Flash as a general-purpose model with a large context window and support for text, images, video and audio input. Its documentation lists a maximum input size of 1,048,576 tokens and an output limit of 65,536 tokens. Those limits are technical capacity figures, not a promise that every application will use the full window efficiently.

The model supports several tools through the Gemini API. Google lists function calling, code execution, search grounding and a computer-use preview among the available capabilities. These tools let an application ask the model to call a defined service, run code in a controlled environment or use external information, but developers remain responsible for permissions, data handling and the checks applied to generated output.

Google is also offering a cybersecurity variant called Gemini 3.8 Flash Cyber. The company says it is intended for trusted defenders and tasks such as threat analysis. Ars Technica reported that the two variants were launched together, giving developers a standard Flash model and a security-focused option rather than one model with a separate safety mode.

Pricing is part of the appeal for high-volume workloads. Google lists an introductory price of $0.75 per million input tokens and $3.75 per million output tokens, matching the launch pricing of the previous Flash generation. The documentation says the introductory pricing is scheduled through December 31, 2026, with a price change listed for January 1, 2027.

Gemini 3.8 Flash is available through the Gemini API and AI Studio, with Google also listing support in Android Studio and selected consumer products. GitHub has separately announced the model for Copilot, with availability rolling out gradually. The release gives developers another option for long-context work, but benchmark claims should be read as vendor or test-specific results rather than a universal ranking.