Exploring Spring AI — Ollama & Stability AI

4 min readDec 25, 2024

Background

Spring community recently released the Spring AI module to integrate the applications with AI models natively with minimal efforts.

Spring has provided the support to all major AI players — https://docs.spring.io/spring-ai/reference/api/index.html

It has also given the space for developers to expose their own models, use different vector stores and leverage dynamic client function calls.

Use Cases

I have tried to cover the below two major use cases -

Chat Completion → Ollama model
Text to Image → Stability AI — stable diffusion model

1. Chat Completion

This feature leverages pre-trained language models, such as GPT (Generative Pre-trained Transformer), to generate human-like responses to user inputs in natural language.

I have pulled the open-source Ollama LLM — llama3.2 model which is lightweight in nature.

There are always better choices coming up. You can look them up here — https://ollama.com/library

Prerequisite Step -> Make sure the Ollama runs in your local machine — Default port: 11434

Spring boot’s Auto-Configuration dependency for Ollama

<!--    Chat Model - Ollama AI -->
<dependency>
    <groupId>org.springframework.ai</groupId>
    <artifactId>spring-ai-ollama-spring-boot-starter</artifactId>
</dependency>

Inject the application properties with the model props

# Ollama properties
#spring.ai.ollama.base-url=localhost:11434 --> incase you wish to customize
spring.ai.ollama.chat.enabled=true
spring.ai.ollama.chat.options.model=llama3.2:1b

Controller Endpoints

@GetMapping("/ask-ai")
public String getOllamaResponse(@RequestParam  String prompt){
    return ollamaChatService.getChatServiceResponse(prompt);
}

@GetMapping("/ask-ai-options")
public String getOllamaResponseOptions(@RequestParam  String prompt){
    return ollamaChatService.getChatServiceResponseOptions(prompt);
}

Service Class

@Service
public class OllamaChatService {

    private final OllamaChatModel ollamaChatModel;

    public OllamaChatService(OllamaChatModel ollamaChatModel) {
        this.ollamaChatModel = ollamaChatModel;
    }

    public String getChatServiceResponse(String prompt){
        return ollamaChatModel.call(prompt);
    }

    public String getChatServiceResponseOptions(String prompt){
        ChatResponse response = ollamaChatModel.call(
                new Prompt(
                        prompt,
                        OllamaOptions.builder()
                                .withModel(OllamaModel.LLAMA3_2_1B)
                                .withTemperature(0.4)
                                .build()
                ));
        return response.getResult().getOutput().getContent();
    }

}

API Response

2. Text To Image

Spring’s auto configuration lets our application access the Stability AI API which generates images from text.

Stability uses stable-diffusion-v1–6 model for image generation (by default)

Spring boot’s Auto-Configuration dependency for Stability AI

<!--    Image Model - Stability AI -->
<dependency>
    <groupId>org.springframework.ai</groupId>
    <artifactId>spring-ai-stability-ai-spring-boot-starter</artifactId>
</dependency>

Inject the application properties with the model props

# Stability properties
#spring.ai.stabilityai.base-url=https://api.stability.ai/v2beta/stable-image/generate/core
#spring.ai.stabilityai.base-url=https://api.stability.ai/v2beta/stable-image/generate/core
spring.ai.stabilityai.api-key=<REPLACE_API_KEY>

spring.ai.stabilityai.image.enabled=true
#spring.ai.stabilityai.image.base-url=https://api.stability.ai/v2beta/stable-image/generate/core
spring.ai.stabilityai.image.api-key=<REPLACE_API_KEY>

spring.ai.stabilityai.image.option.responseFormat="image/png"
#spring.ai.stabilityai.image.option.model=

Stability API Response

Controller Endpoints

@GetMapping("/generate-image-with-stability")
    public List<String> generateStabilityImage(HttpServletResponse response,
                                               @RequestParam String prompt,
                                               @RequestParam(defaultValue = "hd") String quality,
                                               @RequestParam(defaultValue = "1") int n,
                                               @RequestParam(defaultValue = "1024") int width,
                                               @RequestParam(defaultValue = "1024") int height) throws IOException {

        ImageResponse imageResponse = stabilityImageService.generateImage(prompt, quality, n, width, height);

//        String B64Json_Image = imageResponse.getResult().getOutput().getB64Json();
//        byte[] imageBytes = ImageUtils.Base64StringToByteArray(B64Json_Image);
//        System.out.println("Image bytes --> "+ imageBytes.toString());

//        HttpHeaders headers = new HttpHeaders();
//        headers.add("Content-Type", "image/*");
//        headers.add("Content-Disposition", "inline; filename=image.png");


        // Streams to get base 64 jsons from ImageResponse
        List<String> image_jsons = imageResponse.getResults().stream()
                .map(result -> result.getOutput().getB64Json())
                .toList();

        return image_jsons;
    }

Service Class

@Service
public class StabilityImageService {

    private final StabilityAiImageModel stabilityAiImageModel;

    public StabilityImageService(StabilityAiImageModel stabilityAiImageModel) {
        this.stabilityAiImageModel = stabilityAiImageModel;
    }

    public ImageResponse generateImage(String prompt,
                                       String quality,
                                       int n,
                                       int width,
                                       int height)
    {
        ImageResponse imageResponse = stabilityAiImageModel.call(
                new ImagePrompt(prompt,
                        StabilityAiImageOptions.builder()
                        .withStylePreset(StyleEnum.CINEMATIC)
                        .withN(n)
                        .withHeight(height)
                        .withWidth(width).build())
        );

        return imageResponse;
    }

}