Multi Modal Generation

This document shows you how to generate responses and manage the generations.

GEThttps://api.ailibrary.ai/v1/models

List Available Models

This list the models available in the AI Library platform.

There are multiple models available in the platform for different types of data - text, image, and speech. The models are provided by different providers like OpenAI, Anthropic, Google, Stability, and AI Library.

Optional attributes

Name
provider
Type
string
Required
Description
Filter the models by provider. Available providers are openai, anthropic, google, stability, ailibrary

Request

POST

/v1/generate

curl -G https://api.ailibrary.ai/v1/models \
  -H "X-Library-Key: your-api-key"

Response

[
   {
      "modelId": "gpt-4-turbo@openai",
      "name": "GPT-4 Turbo",
      "model": "gpt-4-turbo",
      "provider": "OpenAI",
      "logo": "https://ailib-public.s3.us-west-2.amazonaws.com/public/openai.png",
      "type": "text-to-text"
  },
  ...
]

POSThttps://api.ailibrary.ai/v2/generate

Generate Response

This endpoint allows you to generate responses from multiple AI models and specify knowledge base for Retrieval Augmented Generation (RAG). All responses are in markdown format. In case of image and audio, the format of the response is ![Alt-text](https://example-link.com).

Try in Postman

Request body

Name
modelId
Type
string
Required
required
Description
Unique identifier of the model in the format model@provider. Check the list of available models, here
Name
prompt
Type
string
Required
required
Description
Your prompt
Name
knowledgeId
Type
string
Required
Description
Add the knowledge id for knowledge retrieval.

Response body

Name
text
Type
markdown
Required
Description
Response is generated in markdown format. Images are also available in this field. Image format is ![Alt-text](https://example-link.com).
Name
audio
Type
string
Required
Description
Generated audio is available as a link and can be embedded directly in your application.
Name
context
Type
json
Required
Description
Context of the prompt is available in raw json format. The format is variable and depends on the kind of knowledge base. This is useful for debugging and understanding the context of the prompt.

Request

POST

/v2/generate

import requests
import json

url = "https://api.ailibrary.ai/v2/generate"

payload = json.dumps({
  "modelId": "model@provider",
  "prompt": "your-prompt",
  "urls": [
      "file-id" // optional
  ],
  "knowledge": [
    {
      "id": "knowledge-id"
    }
  ]
})
headers = {
    'X-Library-Key': '••••••',
    'Content-Type': 'application/json'
}

response = requests.request("POST", url, headers=headers, data=payload)

print(response.text)

Response

{
  "text": "Generated text in markdown format from the model",
  "audio": "",
  "provider": "openai",
  "model": "gpt-3.5-turbo-instruct",
  "prompt": "US presidential elections 2024",
  "knowledgeId": "news",
  "context": [
      "... context of the prompt...",
  ],
  "tokens": 1087,
  "userName": "Arani Chaudhuri",
  "userEmail": "arani@ailibrary.ai"
}