Spring AI with OpenAI Models

Muneeb T Khawaja -

See sb3-ai-sample for a sample project based on this post.

Pre-requisites

OpenAI API Key

  • Create an OpenAI Account if you don't already have one.
  • Navigate to the API Keys page and create a new API key.
OpenAI API Keys (https://platform.openai.com/api-keys)
  • Save this key for later as it'll be used to set the SPRING_AI_OPENAI_API_KEY environment variable.

Note: You will be eligible for some free credits ($5) if you created a new account. However, those credits are only valid for 3 months. Consequently, you'll need to purchase some credits if you want to use the API for longer. Review the Pricing - OpenAI and Rate Limits - OpenAI for more information.

Initialize a new Spring AI project

Use the spring-boot cli to create a new spring-ai project:

#!/usr/bin/env bash

spring boot new ai \
  --name . \
  --group-id "com.muneebkhawaja.ai" \
  --artifact-id "sample" \
  --version "0.0.1-SNAPSHOT"

Alternatively, if using PowerShell:

#!/usr/bin/env pwsh

spring.exe boot new ai `
  --name . `
  --group-id "com.muneebkhawaja.ai" `
  --artifact-id "sample" `
  --version "0.0.1-SNAPSHOT"

Note: Update the --name <name-of-project> option if you want to create a new directory for the project.

Your project should have a similar structure to the following:

├── .gitignore
├── README.md
├── mvnw
├── mvnw.cmd
├── pom.xml
└── src
    └── main
        └── java
            └── com
                └── muneebkhawaja
                    └── ai
                        └── sample
                            ├── Application.java
                            └── simple
                                ├── Completion.java
                                └── SimpleAiController.java

Updating Dependencies

At the time of writing, the following dependency is added to the pom.xml when creating the project via the spring cli i.e.


<dependency>
    <groupId>org.springframework.experimental.ai</groupId>
    <artifactId>spring-ai-openai-spring-boot-starter</artifactId>
    <version>0.2.0-SNAPSHOT</version>
</dependency>

According to the spring-ai project README.md on GitHub, the project's group ids are transitioning from org.springframework.experimental.ai to org.springframework.ai. As such, review the Spring AI Documentation for the latest version and update the spring-ai-openai-spring-boot-starter dependency in the pom.xml accordingly i.e.


<dependency>
    <groupId>org.springframework.ai</groupId>
    <artifactId>spring-ai-openai-spring-boot-starter</artifactId>
    <version>0.8.0-SNAPSHOT</version>
</dependency>

Running the application

#!/usr/bin/env bash

export SPRING_AI_OPENAI_API_KEY=<INSERT KEY HERE>
./mvnw spring-boot:run

Alternatively, if using PowerShell:

#!/usr/bin/env pwsh

set-content env:SPRING_AI_OPENAI_API_KEY "<INSERT KEY HERE>"
.\mvnw.cmd  spring-boot:run

The application should be accessible at http://localhost:8080

Trying out endpoints

The quick start project comes with some default endpoints that you can try out i.e.

#!/usr/bin/env bash

curl --get --data-urlencode 'message=Tell me a joke about a cow.' http://localhost:8080/ai/simple

Alternatively, if using PowerShell:

#!/usr/bin/env pwsh

Invoke-WebRequest -Method Get -Uri http://localhost:8080/ai/simple -Body @{ message = "Tell me a joke about a cow." }

A sample response is given below:

Why did the cow go to space?

Because it wanted to see the mooooon!

Notes about Model Usage

The default implementation of AiClient in spring-ai-openai-spring-boot-starter i.e. OpenAiClient
uses the gpt-3.5-turbo model by default.

References