API Reference#

Welcome to the Endpoints API reference! This page is your go-to resource when it comes to learning about the different Unify API endpoints you can interact with.

Note

If you don’t have one yet, Sign Up first to get your API key.

GET /get_credits#

Get Current Credit Balance

Retrieve the credit balance for the authenticated account.

Example Request (curl)

curl -X 'GET' \
  'https://api.unify.ai/v0/get_credits' \
  -H 'accept: application/json' \
  -H 'Authorization: Bearer YOUR_API_KEY'

Responses

200 OK

Successful operation.

Response

Credits balance in the account associated with the API key used for the request.

Example Response
```
{
  "id": "corresponding_user_id",
  "credits": 232.32
}
```
401 Unauthorized

Invalid API key.

Example Response
```
{
  "error": "Invalid API key"
}
```
403 Forbidden

Not authenticated.

Example Response
```
{
  "detail": "Not authenticated"
}
```

POST /inference#

Query a Model hosted in a given Provider

Send a given input to the specified model hosted in the specified provider. Both the arguments and the response are model-specific and, therefore, their format is expected to change from model to model. You can find the specific arguments and the response format in the corresponding model documentation.

For Text-Generation models, you might want to use the POST /chat/completions endpoint.

Request Body: model (string): ID of the model to query.

provider (string): ID of the provider to query.

arguments (object): Model-specific parameters. Check out the model documentation for more information.

Example Request (curl)

curl -X POST "https://api.unify.ai/v0/inference" \
    -H "accept: application/json" \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
        "model": "<model_name>",
        "provider": "<provider_name>",
        "arguments": <Model Inference Input Arguments>
    }'

Responses

200 OK

Successful operation.

Response

Model-specific response, check out the model documentation for more information.

Example Response
```
{
  "response": <Model Inference Output Format>
}
```
401 Unauthorized

Invalid API key.

Example Response
```
{
  "error": "Invalid API key"
}
```
422 Unprocessable Entity

Invalid arguments. The provided arguments don’t correspond to the specified model.

Example Response
```
{
  "error": "The provided arguments don't correspond to the specified model."
}
```

POST /chat/completions#

Query a Text-Generation Model hosted in a given Provider using the OpenAI API format

Send a given input to the specified model hosted in the specified provider. This endpoint follows the OpenAI specification for text completion, which is available here.

To specify the provider, make sure to append its name after the model id using @.

Example Request (curl)

curl -X 'POST' \
'https://api.unify.ai/v0/chat/completions' \
-H 'accept: application/json' \
-H 'Authorization: Bearer YOUR_API_KEY' \
-H 'Content-Type: application/json' \
-d '{
"model": "llama-2-7b-chat@anyscale",
"messages": [
    {
        "role": "user",
        "content": "Explain who Newton was and his entire theory of gravitation. Give a long detailed response please and explain all of his achievements"
    }
],
"stream": false
}'

Responses

200 OK

Successful operation.

Response: Response following the schema of the chat completion object from OpenAI, defined here.

Example Response

{
    'model': 'llama-2-7b-chat@anyscale',
    'created': 1704999905,
    'id': 'meta-llama/Llama-2-7b-chat-hf-xR868C-T4Z-TKLtfXxZSvq57WmhxB34El5ZUuXsAtFU',
    'object': 'chat.completion',
    'usage': {
        'completion_tokens': 512,
        'prompt_tokens': 34,
        'total_tokens': 546
        },
    'choices': [{
        'finish_reason': 'length',
        'index': 0,
        'message': {
            'content': 'Isaac Newton (1643-1727) was a...',
            'role': 'assistant'
        }
    }]
}

401 Unauthorized

Invalid API key.

Example Response
```
{
  "error": "Invalid API key"
}
```
422 Unprocessable Entity

Invalid arguments. The provided arguments don’t correspond to the specified model.

Example Response
```
{
  "error": "The provided arguments don't correspond to the specified model."
}
```