Prompt Optimization - 302.AI API Document

Interface responses are consistent with OpenAI chat model responses, supporting streaming returns.

Optimization methods list:

ID	Name	Tip
1	CRISPE	Optimized according to the LangGPT framework, breaking down simple tasks into complex workflows and generating structured prompts.
2	CO-STAR	Singapore Prompt Competition champion framework, which breaks down, analyzes, and generates structured prompts for simple tasks with one click.
3	Meta Prompting	Prompt optimization method proposed by Tsinghua University and Shanghai AI Laboratory.
4	CoT	Improves the quality and relevance of model-generated content by simulating the thought process of problem-solving.
5	VARI	Google's Deepmind latest research, enhancing prompts through variational planning.
6	Q*	Optimizes prompts using Markov decision processes.
7	RISE	Carnegie Mellon University's latest research, allowing prompts to introspect recursively.
8	O1-STYLE	Mimics O1 by following structured thinking, step-by-step reasoning, continuous reflection, and adjustment strategies in prompts.
9	MicrOptimization	Microsoft's latest research, enhancing your prompt capabilities by automatically optimizing your instruction dataset.
10	OpenAI	OpenAI's officially open-sourced prompt optimization method.
11	claude	Claude's officially open-sourced prompt optimization method.
12	DRAW	Breaks down user input descriptions into six elements: shot, lighting, subject, background, style, and atmosphere, to supplement and refine, generating high-quality drawing prompts.
13	Complete Guide	Manually optimize prompts step by step.
14	modify	Manually optimize prompts.

Price: Calculated based on the model tokens used

Request

Header Params

Authorization

string

required

Example:

Bearer {{YOUR_API_KEY}}

Body Params application/json

integer

优化的方法id

required

Optimization method ID
For specific information, see the table in the document description.

prompt

string

需要优化的提示词

required

Prompt to be optimized

model

string

模型

required

Model

stream

boolean

是否流式返回

optional

Whether to return in streaming mode

Default:

false

modify_value

string

修改要求

optional

Modification requirements
It is mandatory to provide this when selecting manual optimization mode.

Example

{
  "id": 1,
  "prompt": "Help me create some metaphors to describe a person's joyful and infectious laughter.",
  "model": "gpt-4o-mini",
  "stream": false
}

Request samples

Shell

JavaScript

Java

Swift

PHP

Python

HTTP

Objective-C

Ruby

OCaml

Dart

Responses

🟢200成功

application/json

Body

string

required

object

string

required

created

integer

required

model

string

required

choices

array [object {4}]

required

index

integer

optional

message

object

optional

logprobs

null

optional

finish_reason

string

optional

usage

object

required

prompt_tokens

integer

required

completion_tokens

integer

required

total_tokens

integer

required

prompt_tokens_details

object

required

completion_tokens_details

object

required

service_tier

string

required

system_fingerprint

string

required

Example

{
  "id": "string",
  "object": "string",
  "created": 0,
  "model": "string",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "string",
        "content": "string",
        "refusal": null,
        "annotations": [
          "string"
        ]
      },
      "logprobs": null,
      "finish_reason": "string"
    }
  ],
  "usage": {
    "prompt_tokens": 0,
    "completion_tokens": 0,
    "total_tokens": 0,
    "prompt_tokens_details": {
      "cached_tokens": 0,
      "audio_tokens": 0
    },
    "completion_tokens_details": {
      "reasoning_tokens": 0,
      "audio_tokens": 0,
      "accepted_prediction_tokens": 0,
      "rejected_prediction_tokens": 0
    }
  },
  "service_tier": "string",
  "system_fingerprint": "string"
}