Overview
The Presets API allows you to save and manage conversation configurations for quick reuse. Presets store model settings, parameters, and endpoint configurations.
All preset endpoints are prefixed with /api/presets.
List Presets
Retrieve all presets for the authenticated user:
GET /api/presets
Authorization: Bearer <token>
Response
Returns an array of preset objects:
[
{
"presetId": "preset_abc123",
"title": "GPT-4 Creative Mode",
"endpoint": "openAI",
"model": "gpt-4",
"modelLabel": "GPT-4",
"temperature": 0.9,
"top_p": 1,
"presence_penalty": 0.6,
"frequency_penalty": 0.5,
"max_tokens": 2000,
"instructions": "You are a creative writing assistant...",
"createdAt": "2024-01-15T10:00:00Z",
"updatedAt": "2024-01-15T10:00:00Z"
},
{
"presetId": "preset_xyz789",
"title": "Claude Code Helper",
"endpoint": "anthropic",
"model": "claude-3-opus",
"temperature": 0.3,
"max_tokens": 4096,
"instructions": "You are an expert programmer...",
"createdAt": "2024-01-14T09:00:00Z"
}
]
Create or Update Preset
Create a new preset or update an existing one:
POST /api/presets
Authorization: Bearer <token>
Content-Type: application/json
{
"title": "My Custom Preset",
"endpoint": "openAI",
"model": "gpt-4",
"temperature": 0.7,
"top_p": 1,
"presence_penalty": 0,
"frequency_penalty": 0,
"max_tokens": 1500,
"instructions": "You are a helpful assistant."
}
Request Body
Preset ID (auto-generated if not provided). Include to update existing preset.
AI endpoint (e.g., openAI, anthropic, google, azureOpenAI)
Model identifier (e.g., gpt-4, claude-3-opus, gemini-pro)
Display label for the model
Sampling temperature (0-2). Higher values make output more random.
Nucleus sampling parameter (0-1)
Presence penalty (-2 to 2). Positive values penalize new tokens based on whether they appear in the text so far.
Frequency penalty (-2 to 2). Positive values penalize new tokens based on their frequency in the text so far.
Maximum number of tokens to generate
System instructions/prompt
OpenAI-specific Parameters
Whether to resend files with each message (OpenAI)
Image detail level: auto, low, or high (OpenAI Vision)
Anthropic-specific Parameters
Top-k sampling parameter (Anthropic)
Google-specific Parameters
Safety settings configuration (Google)
Response
Returns the created/updated preset object with HTTP status 201.
{
"presetId": "preset_abc123",
"title": "My Custom Preset",
"endpoint": "openAI",
"model": "gpt-4",
"temperature": 0.7,
"top_p": 1,
"presence_penalty": 0,
"frequency_penalty": 0,
"max_tokens": 1500,
"instructions": "You are a helpful assistant.",
"createdAt": "2024-01-15T10:00:00Z",
"updatedAt": "2024-01-15T10:00:00Z"
}
Delete Preset
Delete one or more presets:
POST /api/presets/delete
Authorization: Bearer <token>
Content-Type: application/json
{
"presetId": "preset_abc123"
}
Request Body
ID of preset to delete. If not provided, all user presets are deleted.
Response
Number of presets deleted
Preset Configuration Examples
Creative Writing
{
"title": "Creative Writer",
"endpoint": "openAI",
"model": "gpt-4",
"temperature": 0.9,
"top_p": 1,
"presence_penalty": 0.6,
"frequency_penalty": 0.5,
"instructions": "You are a creative writing assistant who helps with storytelling, character development, and narrative structure."
}
Code Assistant
{
"title": "Code Helper",
"endpoint": "anthropic",
"model": "claude-3-opus",
"temperature": 0.2,
"top_p": 0.9,
"max_tokens": 4096,
"instructions": "You are an expert programmer. Provide clear, well-documented code with explanations.",
"tools": ["code_interpreter"]
}
Research Assistant
{
"title": "Research Helper",
"endpoint": "google",
"model": "gemini-pro",
"temperature": 0.4,
"topK": 40,
"max_tokens": 2048,
"instructions": "You are a thorough research assistant. Provide accurate, well-cited information.",
"tools": ["web_search"]
}
Concise Responses
{
"title": "Concise Mode",
"endpoint": "openAI",
"model": "gpt-4",
"temperature": 0.5,
"max_tokens": 500,
"instructions": "Provide brief, direct answers. Be concise and to the point."
}
Preset Parameters by Endpoint
OpenAI
Supported parameters:
temperature (0-2)top_p (0-1)presence_penalty (-2 to 2)frequency_penalty (-2 to 2)max_tokensstopresendFilesimageDetail
Anthropic (Claude)
Supported parameters:
temperature (0-1)top_p (0-1)top_k (1-500)max_tokensstop_sequences
Google (Gemini)
Supported parameters:
temperature (0-2)topP (0-1)topK (1-40)maxOutputTokenssafetySettings
Azure OpenAI
Same as OpenAI parameters.
Error Responses
Invalid Preset Data
{
"error": "There was an error when saving the preset",
"message": "Invalid model specified"
}
Preset Not Found
{
"error": "Preset not found"
}
Delete Error
{
"error": "There was an error deleting the presets"
}
Use Cases
Quick Model Switching
Save presets for different models to quickly switch between GPT-4, Claude, and Gemini without reconfiguring.
Task-Specific Configurations
Create presets optimized for specific tasks:
- Code review (low temperature, code tools)
- Creative writing (high temperature, presence penalty)
- Data analysis (code interpreter enabled)
- Research (web search enabled)
Team Collaboration
Share preset configurations with team members by exporting/importing preset JSON.
A/B Testing
Compare different parameter configurations by creating multiple presets and testing which works best for your use case.
TypeScript Types
import type { TPreset } from 'librechat-data-provider';
interface TPreset {
presetId?: string;
title: string;
endpoint: EModelEndpoint;
model: string;
modelLabel?: string;
temperature?: number;
top_p?: number;
presence_penalty?: number;
frequency_penalty?: number;
max_tokens?: number;
instructions?: string;
stop?: string[];
tools?: string[];
agent_id?: string;
// Endpoint-specific parameters
resendFiles?: boolean; // OpenAI
imageDetail?: string; // OpenAI Vision
top_k?: number; // Anthropic
topK?: number; // Google
safetySettings?: object[]; // Google
}