Gemini V SDK

A V language SDK for interacting with Google's Gemini AI API. This module provides a simple and efficient way to integrate Gemini's generative AI capabilities into your V applications.

Features

• Support for multiple free Gemini models (2.5 Flash variants, 3.0 Flash and gemma 3 variant) - PRs welcome for paid models
• Simple prompt-based text generation
• Advanced request configuration with system instructions
• Comprehensive response handling
• Type-safe API with structured data models
• Built-in error handling

Installation

v install Ddiidev.sdkgemini

OR

v install https://github.com/Ddiidev/sdk_gemini

Quick Start

// import Ddiidev.sdk_gemini //for install from vpm
// for install github repo
import log // built-in V module
import sdk_gemini

fn main() {
    //log level 'debug' provides full model response when errors occur
    log.set_level(.debug)

    // Read API key from environment
    api_key := sdk_gemini.get_api_key('GEMINI_API_KEY') or {
        log.error(err.msg())
        return
    }

    // Initialize the SDK with your API key
    mut sdk := sdk_gemini.new(api_key)

    // Send a simple prompt
    response := sdk.send_prompt(
        .gemini_3_0_flash_preview,
        'Tell me about João Pessoa, Paraíba',
        'You are a tour guide from Brazil, specifically from the Northeast region, João Pessoa and Paraíba state'
    ) or {
            log.error(err.msg())
            return
    }

    content := response.str()
    println(content)
}

Safe Way to Get Content

Use the GeminiResponse iterator to safely iterate through all parts.

for part in response {
    println(part)
}

To get all concatenated content in a single string, use str():

text := response.str()
if text != '' {
    println(text)
}

You can also directly get the last part with last():

last := response.last() or { '' }
if last != '' {
    println(last)
}

API Reference

GeminiSDK

The main SDK struct for interacting with the Gemini API.

pub struct GeminiSDK {
pub:
    api_key string
}

Methods

completation(model Models, req_payload GeminiRequest) !GeminiResponse

Sends a complete request to the Gemini API with full control over the request structure.

Parameters:- model: The Gemini model to use (see Models enum)

• req_payload: Complete request configuration

Returns: GeminiResponse or error

send_prompt(model Models, prompt string, system_instruction string) !GeminiResponse

Simplified method for sending text prompts with optional system instructions.

Parameters:- model: The Gemini model to use

• prompt: The user prompt/question
• system_instruction: Optional system instruction (can be empty string)

Returns: GeminiResponse or error

Supported Models

The SDK supports the following Gemini models:

• gemma_3_1b_it - Gemma 3.1b It
• gemma_3_4b_it - Gemma 3.4b It
• gemma_3_12b_it - Gemma 3.12b It
• gemma_3_27b_it - Gemma 3.27b It
• gemini_2_5_flash_lite_preview_09_2025 - Gemini 2.5 Flash Lite Preview 09 2025
• gemini_2_5_flash_lite - Gemini 2.5 Flash Lite
• gemini_2_5_flash - Gemini 2.5 Flash
• gemini_3_0_flash_preview - Gemini 3.0 Flash Preview

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit issues and pull requests.

Links

• Google Gemini API Documentation
• V Language Documentation

fn get_api_key(key_var string) !string

get_api_key Reads an Gemini API key from an environment variable.

fn new(api_key string) GeminiSDK

new Returns a GeminiSDK.

struct GeminiSDK {
pub:
	api_key string
}

fn (mut sdk GeminiSDK) completation(model structs.Models, req_payload structs.GeminiRequest) !structs.GeminiResponse

completation Sends a request to the Gemini API and returns the response or an error. Log level 'debug' logs the full model response body in case of an error occuring.

fn (mut sdk GeminiSDK) send_prompt(model structs.Models, prompt string, system_instruction string) !structs.GeminiResponse

send_prompt Sends a prompt to the Gemini API and returns the model's response.