Kotlin Multiplatform SDK to access Google’s Gemini via Vertex AI on Android and iOS

This SDK provides a simple and idiomatic way to integrate Google’s Gemini (powered by Vertex AI) into your Android and iOS applications using Kotlin Multiplatform.

• A Compose Multiplatform project with either an Android or iOS target
• Kotlin 2.1.10 or higher

1. Follow the instructions at this link to set up your Google Cloud project: Setup GCP.
2. Follow the instructions at this link to set up your local environment: Setup Local Environment.
3. Run the following command in your CLI to generate an access token and make note of this key:

   gcloud auth print-access-token
   

Use the generated key in your application wherever the API_KEY is required.

// composeApp/build.gradle.kts
kotlin {
    commonMain.dependencies {
        implementation(libs.vertexgsoc)
    }
}

// gradle/libs.versions.toml
[versions]
vertexgsoc = "1.0.0"

[libraries]
vertexgsoc = { group = "io.github.seanchinjunkai", name = "vertexgsoc", version.ref = "vertexgsoc" }

val model =  GenerativeModel(key = "API_KEY", model = "gemini-1.5-flash-002", RequestOptions(projectId = "gemini-kmp"))

val response: String? = result.response?.text 

Try these apps built with this SDK by the community:

• SeanChinJunKai/VertexSample: Kotlin Multiplatform sample that uses Gemini via Vertex AI. Runs on Android and iOS.

This library is currently in its initial development stages and is not intended for production use. The following reasons should be considered:

1. Limited Features: Not all features of Google’s Gemini via Vertex AI are supported, as outlined in the features table above.
2. Lack of Production-Grade Stability: The SDK has not been extensively tested for large-scale production environments, and the codebase may undergo significant changes in the future.
3. Security Considerations: As this library handles sensitive data such as API keys, proper security and compliance measures may not be fully implemented or sufficient for production systems.
4. Performance Constraints: The performance of this SDK in high-load or real-time production scenarios has not been benchmarked.

If you choose to use this library, it is highly recommended to do so for prototyping, experimentation, or non-critical applications only. For production-grade solutions, consider implementing or integrating rigorously tested alternatives or tools.