Integrate Magikfake with your product via our APIs
To use Magikfake's API, users must authenticate their requests using their API key.
The level of access and usage limits for the user are determined by the subscription plan they choose.
By limiting usage based on the selected plan, Magikfake enables users to choose the plan that best suits their needs and budget. This allows for greater flexibility and cost-effectiveness, as users only pay for the level of access they require.
In addition to using the API directly, users can also interface with the Magikfake system using the provided npm SDK. This SDK provides a range of helpful tools and utilities for working with the Magikfake API, making it easier and more efficient to generate fake data for a variety of applications and use cases.
| Base url |
|---|
| Plan | Limits |
|---|---|
| Free | 25.000 tokens/monthMax records number: 20 |
The consumption of tokens in an API call or an interaction with a language model is determined by several key parameters:
Standard Consumption for Initial Analysis (Overhead):
A fixed amount of tokens is consumed for the preliminary context analysis.
By default, the call consumes between 3000 and 3500 standard tokens to determine which datasets or data sources to use and how to proceed with information retrieval.
User Argument Length:
The length of the argument provided by the user (the prompt or query) directly contributes to the token count.
The longer and more complex the argument, the higher the consumption.
Output Schema (JSON Schema):
If the user defines a JSON schema for the desired output, its length and complexity are considered in the token calculation.
The schema guides the model and thus influences its 'thought process' and consumption.
Reasoning Tokens ('Thought Tokens'):
A number of thought tokens (or 'reasoning tokens') are consumed to allow the model to perform the correct mapping, plan the response, and select the appropriate information before generating the output.
What is not considered:
Data Record Length (JSON Payload):
The data that the model retrieves or generates and presents to the user in JSON format (e.g., a list of records, query results, etc.) is not counted as part of the consumption