A simple endpoint to check if the Prediction Guard API is healthy.

example1

Check API Health

A simple endpoint to check what models are available in the /chat/completions endpoint.

List Chat Models

Generate chat completions based on a conversation history.

A basic example of using the API.

The chat model to use for generating completions.

An array of messages used for generating completions.

The maximum number of tokens in the generated completion.

The temperature parameter for controlling randomness in completions.

The diversity of the generated text based on nucleus sampling.

The diversity of the generated text based on top-k sampling.

Options to affect the output of the response.

Options to affect the input of the request.

Chat Completions

Generate chat completions based on a conversation history with streaming support.

Chat Completions Streaming

Generate chat completions based on text and images.

Chat Completions Vision

A simple endpoint to check what models are available in the /completions endpoint.

List Completions Models

Retrieve text completions based on the provided input.

The prompt to use for generating completions.

Completions

A simple endpoint to check what models are available in the /embeddings endpoint.

List Embeddings Models

Generate vector data points from text and images.

Embeddings

Check the factuality of a given text compared to a reference.

The president of the united states can take a salary of one million dollars

Factuality

Injection detects potential prompt injection attacks.

The prompt to check for injection attack.

Whether to detect potential injection attacks.

Injection

Replace personal information such as names, SSNs, and emails in a given text.

The prompt to check for personal information.

The method to use (random, fake, category, mask).

Replace Personal Information

Every flight I have is late and I am very angry. I want to hurt someone.

Toxicity

Translate text from one language to another.

Enables/disables third party translations.

Translate

PostChatCompletionsRequestBadRequestErrorBody

PostChatCompletionsRequestForbiddenErrorBody

PostChatCompletionsRequestInput

Set to true to detect prompt injection attacks.

Set to either 'random', 'fake', 'category', 'mask'.

PostChatCompletionsRequestMessagesItem

The role of the sender (user or assistant).

PostChatCompletionsRequestOutput

Set to true to turn on factuality processing.

Set to true to turn on toxicity processing.

PostChatCompletionsResponse

Timestamp of when the chat completion was created.

The chat model used for generating completions.

PostChatCompletionsResponseChoicesItem

The response if this choice was successful.

PostChatCompletionsResponseChoicesItemMessage

PostChatCompletionsStreamingVersionRequestBadRequestErrorBody

PostChatCompletionsStreamingVersionRequestForbiddenErrorBody

PostChatCompletionsStreamingVersionRequestInput

PostChatCompletionsStreamingVersionRequestMessagesItem

PostChatCompletionsStreamingVersionResponseItem

PostChatCompletionsStreamingVersionResponseItemData

PostChatCompletionsStreamingVersionResponseItemDataChoicesItem

The final completed chat response which is provided when this is the last choice.

The log probabilty of accuracy for this choice.

PostChatCompletionsStreamingVersionResponseItemDataChoicesItemDelta

The partial content response for a choice.

PostChatCompletionsVisionVersionRequestBadRequestErrorBody

PostChatCompletionsVisionVersionRequestForbiddenErrorBody

PostChatCompletionsVisionVersionRequestInput

PostChatCompletionsVisionVersionRequestMessagesItem

PostChatCompletionsVisionVersionRequestMessagesItemContentItem

PostChatCompletionsVisionVersionRequestMessagesItemContentItemImageUrl

PostChatCompletionsVisionVersionRequestOutput

PostChatCompletionsVisionVersionResponse

PostChatCompletionsVisionVersionResponseChoicesItem

PostChatCompletionsVisionVersionResponseChoicesItemMessage

PostCompletionsRequestBadRequestErrorBody

PostCompletionsRequestForbiddenErrorBody

PostCompletionsRequestInput

PostCompletionsRequestOutput

PostCompletionsResponse

PostCompletionsResponseChoicesItem

The model used for generating the result.

PostEmbeddingsRequestBadRequestErrorBody

PostEmbeddingsRequestForbiddenErrorBody

PostEmbeddingsRequestInputItem

The text to vectorize. You can choose to provide this or not.

The base64 encoding of an image. You can choose to provide this or not.

PostEmbeddingsResponse

PostEmbeddingsResponseDataItem

PostFactualityRequestBadRequestErrorBody

PostFactualityRequestForbiddenErrorBody

PostFactualityResponse

PostFactualityResponseChecksItem

PostInjectionRequestBadRequestErrorBody

PostInjectionRequestForbiddenErrorBody

PostInjectionResponse

PostInjectionResponseChecksItem

The probability of a potential injection attack.

PostPiiRequestBadRequestErrorBody

PostPiiRequestForbiddenErrorBody

PostPiiResponse

PostPiiResponseChecksItem

The new prompt with the replaced information.

PostToxicityRequestBadRequestErrorBody

PostToxicityRequestForbiddenErrorBody

PostToxicityResponse

PostToxicityResponseChecksItem

PostTranslateRequestBadRequestErrorBody

PostTranslateRequestForbiddenErrorBody

PostTranslateResponse

The set translations from different models.

PostTranslateResponseTranslationsItem

The model that was used for this translation.

The status of using the model for this translation.

1	curl -X POST https://api.predictionguard.com/toxicity \
2	-H "x-api-key: <apiKey>" \
3	-H "Content-Type: application/json" \
4	-d '{
5	"text": "Every flight I have is late and I am very angry. I want to hurt someone."
6	}'

1	{
2	"id": "toxi-bcUAQEaWhMweS3Jpi3nT9SJys28XN",
3	"object": "toxicity_check",
4	"created": 1717676257,
5	"checks": [
6	{
7	"index": 0,
8	"score": 0.7072361707687378,
9	"status": "success"
10	}
11	]
12	}

Toxicity

Request

Response

Errors