Text Completion API

Authentication header. Uses a Bearer token, e.g. Bearer YOUR_API_KEY.

Request content type, must be application/json.

Model ID, e.g. gpt-3.5-turbo-instruct. You can check available models via the Model List.

The prompt(s) to generate completions for. Can be a string, an array of strings, an array of tokens, or an array of token arrays. Officially defaults to <|endoftext|> (the model generates as if from the beginning of a new document).

The suffix that comes after a completion of inserted text. Only supported for gpt-3.5-turbo-instruct.

The maximum number of tokens to generate, officially defaults to 16. The token count of your prompt plus max_tokens cannot exceed the model's context length.

Sampling temperature, between 0 and 2, officially defaults to 1. Higher values make output more random, lower values more focused and deterministic. We generally recommend altering this or top_p but not both.

Nucleus sampling parameter, between 0 and 1, officially defaults to 1.

How many completions to generate for each prompt, between 1 and 128, officially defaults to 1. This can quickly consume your token quota; use reasonable max_tokens and stop settings.

Whether to stream back partial progress, officially defaults to false. When enabled, tokens are sent as server-sent events as they become available, terminated by data: [DONE].

Additional options for streaming output. Only set this when stream: true.

Include the log probabilities on the logprobs most likely tokens, as well as the chosen tokens. Maximum value is 5; the response may contain up to logprobs + 1 elements.

Whether to echo back the prompt in addition to the completion, officially defaults to false.

Up to 4 sequences where the API will stop generating further tokens. The returned text will not contain the stop sequence.

Number between -2.0 and 2.0, officially defaults to 0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the likelihood of talking about new topics.

Number between -2.0 and 2.0, officially defaults to 0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the likelihood of repeating the same line verbatim.

Generates best_of completions server-side and returns the one with the highest log probability per token. Officially defaults to 1, maximum 20. Cannot be streamed; when used with n, best_of must be greater than n. Use carefully as it can quickly consume your token quota.

Modifies the likelihood of specified tokens appearing in the completion. A JSON object mapping token IDs (in the GPT tokenizer) to bias values from -100 to 100. For example, passing {"50256": -100} prevents the <|endoftext|> token from being generated.

Random seed. With the same seed and parameters, the system makes a best effort to sample deterministically, though determinism is not guaranteed; monitor backend changes via the system_fingerprint response field.

A unique identifier representing your end-user, which can help monitor and detect abuse.

The completion response ID.

Object type, always text_completion.

The timestamp when the response was created.

The model that generated the response.

Array of candidate results.

Token usage statistics.