This documentation covers the backend_api.py module in the GPT4Free project, which serves as the core backend API handler for the web interface. It provides endpoints for interacting with language models, managing files, handling conversations, and processing media.
GET /backend-api/v2/models
Returns a list of all available models with their capabilities (image, vision, audio, video) and compatible providers.
GET /backend-api/v2/models/<provider>
Returns models available for a specific provider.
GET /backend-api/v2/providers
Returns a list of all available providers with their capabilities and authentication requirements.
GET /backend-api/v2/version
Returns current and latest version information.
POST /backend-api/v2/conversation
Handles conversation requests and streams responses. Supports text, images, and other media.
Request Body:
{
"model": "gpt-3.5-turbo",
"provider": "OpenAI",
"messages": [
{"role": "user", "content": "Hello, how are you?"}
],
"download_media": true
}GET /backend-api/v2/create
Creates a simple conversation with optional web search integration.
Query Parameters:
-
model: Model to use -
prompt: User prompt -
provider: Provider to use -
web_search: Enable web search (true/false or search query) -
filter_markdown: Filter markdown in response
POST /backend-api/v2/usage
Logs usage data for analytics.
POST /backend-api/v2/log
Logs client-side events.
GET/DELETE /backend-api/v2/files/<bucket_id>
Retrieves or deletes file buckets.
POST /backend-api/v2/files/<bucket_id>
Uploads files to a bucket. Supports various file types including documents and media.
GET /files/<bucket_id>/media/<filename>
Retrieves media files from a bucket.
GET /search/<search>
Searches for media files based on search terms.
POST /backend-api/v2/upload_cookies
Uploads cookie files for authentication with providers.
GET /backend-api/v2/chat/<share_id>
Retrieves a shared chat by ID.
POST /backend-api/v2/chat/<share_id>
Uploads or updates a shared chat.
POST /backend-api/v2/memory/<user_id>
Adds memory items for a user.
GET /backend-api/v2/memory/<user_id>
Retrieves memory items for a user with optional search and filtering.
GET /images/<path:name>
GET /media/<path:name>
Serves image and media files.
GET /backend-api/v2/synthesize/<provider>
Handles audio synthesis requests for text-to-speech functionality.
// Example using fetch API
fetch('/backend-api/v2/conversation', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'gpt-3.5-turbo',
provider: 'OpenAI',
messages: [
{role: 'user', content: 'What is artificial intelligence?'}
]
})
})
.then(response => {
// Handle streaming response
const reader = response.body.getReader();
// Process chunks...
})// Example using FormData
const formData = new FormData();
formData.append('files', fileObject);
fetch('/backend-api/v2/files/my-bucket', {
method: 'POST',
body: formData
})
.then(response => response.json())
.then(data => console.log(data));fetch('/backend-api/v2/create?model=gpt-3.5-turbo&prompt=Latest news about AI&web_search=true', {
method: 'GET'
})
.then(response => {
// Handle streaming response with web search results
})The API uses a consistent JSON format for responses with a type field indicating the kind of response:
-
content: Text content from the model -
provider: Information about the selected provider -
conversation: Conversation state -
preview: Preview of content -
error: Error information -
auth: Authentication requirements -
parameters: Model parameters -
finish: Finish reason -
usage: Token usage information -
reasoning: Reasoning process -
suggestions: Suggested follow-up questions -
log: Debug logs
- The API supports streaming responses for real-time interaction
- File uploads can be automatically converted to markdown if MarkItDown is available
- Media files are stored in a dedicated directory structure with caching
- Web search functionality enhances responses with real-world information
- The system includes caching mechanisms for improved performance
This documentation provides an overview of the main functionality in the Backend API. For more detailed information, refer to the code or reach out to the development team.