Modus enables you to easily integrate AI models into your app. In just a few steps, you can generate text, classify items, compute embeddings, and use models in your app for many other use cases using the models API in the Modus SDK.

​

Understanding key components

Models: your app can invoke models hosted on Hypermode, OpenAI, Anthropic, and many more. You define models in your app manifest.

Models API: the models API in the Modus SDK provides a set of functions that you can import and call from your app.

​

Define your models

You define models in your app manifest. Here are some examples:

{
  ...
  "models": {
    // model card: https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct
    "text-generator": {
      "sourceModel": "meta-llama/Llama-3.2-3B-Instruct", // model name on the provider
      "provider": "hugging-face", // provider for this model
      "connection": "hypermode" // host where the model is running
    }
  }
  ...
}

​

Invoking a model for inference

To invoke a model within your app, import the models packages from the SDK. Import the core models package and the package for the interface your model uses. For example, to use the OpenAI interface for a text-generation model, you would import the openai package in addition to the core models package.

​

Generation models

Generation models are models that generate text, images, or other data based on input. Currently, the Models API supports the OpenAI, Anthropic, and Gemini interfaces. Let’s see how to invoke a model using the OpenAI interface.

When using a model interface, you automatically get type-ahead guidance in your code editor based on the available options for that interface.

package main

import (
    "encoding/json"
    "fmt"
    "strings"

    "github.com/hypermodeinc/modus/sdk/go/pkg/models"
    "github.com/hypermodeinc/modus/sdk/go/pkg/models/openai"
)

// this model name should match the one defined in the modus.json manifest file
const modelName = "text-generator"

func GenerateText(instruction, prompt string) (string, error) {
    model, err := models.GetModel[openai.ChatModel](modelName)
    if err != nil {
        return "", err
    }

    input, err := model.CreateInput(
        openai.NewSystemMessage(instruction),
        openai.NewUserMessage(prompt),
    )
    if err != nil {
        return "", err
    }

    // this is one of many optional parameters available for the OpenAI chat interface
    input.Temperature = 0.7

    output, err := model.Invoke(input)
    if err != nil {
        return "", err
    }

    return strings.TrimSpace(output.Choices[0].Message.Content), nil
}

​

Classification models

Classification models provide a label for input data. You can use these models to sort data into categories or classes. Let’s see how to invoke a classification model.

import (
  "errors"
  "fmt"

  "github.com/hypermodeinc/modus/sdk/go/pkg/models"
  "github.com/hypermodeinc/modus/sdk/go/pkg/models/experimental"
)

// this model name should match the one defined in the modus.json manifest file
const modelName = "my-classifier"

// this function takes input text and a probability threshold, and returns the
// classification label determined by the model, if the confidence is above the
// threshold; otherwise, it returns an empty string
func ClassifyText(text string, threshold float32) (string, error) {
  predictions, err := classify(text)
  if err != nil {
    return "", err
  }

  prediction := predictions[0]
  if prediction.Confidence < threshold {
    return "", nil
  }

  return prediction.Label, nil
}

​

Embedding models

Modus supports invoking embedding models for text, images, and other data types. You use the outputs of these models for implementing search, recommendation, and similarity functions in your app.