API requests
See the API docs.
See sections below for tips on using the text and image endpoints.
Notes:
- The links below will work fine in the browser as
GET
requests but these are intended to handled in a client e.g. using Python or React for building the URL and displaying the content for a user. - If you use Firefox rather than CHrome, the percent encoding is rendered as a space, so that’s easy to read. Or you can type an underscore for a space to keep it readable.
Text endpoint
e.g.
The text endpoint also includes a path for an API that follows the OpenAI specification, so it can be used easily with the openai
Python/JavaScript packages or Python LangChain’s OpenAI connector as covered below.
https://text.pollinations.ai/openai
With the openai package:
from openai import OpenAI
client = OpenAI(base_url="https://text.pollinations.ai/openai", api_key="dummy")
With the langchain-openai package and ChatOpenAI class:
from langchain_openai import ChatOpenAI
client = ChatOpenAI(base_url="https://text.pollinations.ai/openai", openai_api_key="dummy")
Options
- GET -
prompt*
,model
,seed
,json
,system
- POST -
messages*
(can include system prompt),model
,seed
,jsonMode
See explanations in the Text API docs.
See models here: text.pollinations.ai/models.
Image endpoint
Use the form on the Pollinations homepage. There is a feed of other users’ images and you can edit the prompt or make a new one.
A generated image can be viewed there and the link can be copied for ease, rather than building the URL by hand.
e.g. image.pollinations.ai/prompt/calm%20serene%20forest…&nologo=true&model=flux
Options
See Image options in the API docs.
See models here: image.pollinations.ai/models.
Based on this page - pollinations-react/README.md