An Ollama client SDK that allows you to easily interact with the Ollama API.

The supported API follows the Ollama OpenAPI.

• Comprehensive API Coverage: Full support for the Ollama API, including chat, streaming, model management, and embeddings.
• Kotlin Multiplatform: Write once, run on any platform that Kotlin supports.
• Asynchronous Operations: Built with Kotlin Coroutines for non-blocking, asynchronous API calls.
• Type-Safe DSL: A clean and expressive DSL for building requests.
• Streaming Support: Handle large responses efficiently with built-in streaming capabilities.

• Kotlin 1.8 or higher
• JVM 11 or higher

Add the dependency to your Gradle configuration:

implementation("org.nirmato.ollama:nirmato-ollama-client-ktor:0.2.0")

// example using ktor CIO engine
implementation("io.ktor:ktor-client-cio:3.1.3")

or to your Maven pom:

<dependency>
    <groupId>org.nirmato.ollama</groupId>
    <artifactId>nirmato-ollama-client-ktor</artifactId>
    <version>0.2.0</version>
</dependency>

<!-- example using ktor CIO engine -->
<dependency>
    <groupId>io.ktor</groupId>
    <artifactId>ktor-client-cio</artifactId>
    <version>3.1.3</version>
</dependency>

The OllamaClient class contains all the methods needed to interact with the Ollama API. An example of calling Ollama:

val ollamaClient = OllamaClient(CIO) {
    httpClient {
        // ktor HttpClient configurations
        defaultRequest {
            url("http://localhost:11434/api/")
        }
    }
}

val request = chatRequest {
    model("tinyllama")
    messages(listOf(Message(role = USER, content = "Why is the sky blue?")))
}

val response = ollamaClient.chat(request)

See the samples directory for complete examples.

The OllamaApi interface provides the following methods:

• chat(chatRequest): Send a chat request and get a response.
• chatStream(chatRequest): Stream the chat response.
• generateEmbed(generateEmbedRequest): Generate embeddings for a given input.
• createModel(createModelRequest): Create a new model.
• pullModel(pullModelRequest): Pull a model from the registry.
• pushModel(pushModelRequest): Push a model to the registry.
• listModels(): List all available models.
• ...and many more.

For a complete list of methods, please refer to the OllamaApi interface.

Contributions are welcome! Please feel free to submit a Pull Request.

• JetBrains for creating Kotlin.

The source code is distributed under the Apache License 2.0.