Models API

The Models API allows you to list and manage available AI models across different providers. Dispersl supports multiple LLM providers including Anthropic, OpenAI, DeepSeek, and Google.

Overview

Dispersl is model-agnostic, allowing you to:

Use any supported LLM model
Swap providers on the fly
Maintain context across different models
Choose the best model for each task type

Supported Providers

Anthropic: Claude 3 Sonnet, Claude 3 Haiku, Claude 3 Opus
OpenAI: GPT-4, GPT-4 Turbo, GPT-3.5 Turbo
DeepSeek: DeepSeek R1, DeepSeek Coder
Google: Gemini Pro, Gemini Pro Vision

Endpoints

List Available Models

GET /models

Retrieves all available AI models with their specifications and tier requirements.

Headers

Authorization: Bearer YOUR_API_KEY

Response

{
  "models": [
    {
      "id": "anthropic/claude-3-sonnet",
      "name": "Claude 3 Sonnet",
      "description": "Anthropic's most balanced model for complex reasoning and analysis",
      "context_length": 200000,
      "tier_requirements": {
        "free_model": false
      }
    },
    {
      "id": "openai/gpt-4",
      "name": "GPT-4",
      "description": "OpenAI's most capable model for complex multi-step tasks",
      "context_length": 128000,
      "tier_requirements": {
        "free_model": false
      }
    },
    {
      "id": "openai/gpt-3.5-turbo",
      "name": "GPT-3.5 Turbo",
      "description": "Fast and efficient model for simpler tasks",
      "context_length": 16385,
      "tier_requirements": {
        "free_model": true
      }
    },
    {
      "id": "deepseek/deepseek-r1",
      "name": "DeepSeek R1",
      "description": "Advanced reasoning model optimized for code and mathematics",
      "context_length": 65536,
      "tier_requirements": {
        "free_model": false
      }
    },
    {
      "id": "google/gemini-pro",
      "name": "Gemini Pro",
      "description": "Google's multimodal model for text and vision tasks",
      "context_length": 32768,
      "tier_requirements": {
        "free_model": true
      }
    }
  ],
  "status": "success"
}

Model Selection Guide

For Code Generation

Primary: anthropic/claude-3-sonnet - Excellent for complex code architecture
Alternative: deepseek/deepseek-r1 - Specialized for coding tasks
Budget: openai/gpt-3.5-turbo - Good for simple code generation

For Testing

Primary: openai/gpt-4 - Comprehensive test coverage
Alternative: anthropic/claude-3-sonnet - Good test reasoning
Budget: openai/gpt-3.5-turbo - Basic test generation

For Documentation

Primary: google/gemini-pro - Excellent for documentation
Alternative: anthropic/claude-3-sonnet - Clear explanations
Budget: openai/gpt-3.5-turbo - Simple documentation

For Git Operations

Primary: deepseek/deepseek-r1 - Good for version control logic
Alternative: openai/gpt-4 - Reliable for git workflows
Budget: openai/gpt-3.5-turbo - Basic git operations

For Chat/Analysis

Primary: anthropic/claude-3-sonnet - Best for analysis and reasoning
Alternative: openai/gpt-4 - Good conversational abilities
Budget: openai/gpt-3.5-turbo - Quick responses

Context Length Considerations

Model	Context Length	Best For
Claude 3 Sonnet	200,000 tokens	Large codebases, extensive documentation
GPT-4	128,000 tokens	Complex multi-step tasks
DeepSeek R1	65,536 tokens	Code analysis and generation
Gemini Pro	32,768 tokens	Multimodal tasks, documentation
GPT-3.5 Turbo	16,385 tokens	Quick tasks, simple operations

Tier Requirements

Free Tier Models

openai/gpt-3.5-turbo
google/gemini-pro

Pro/Enterprise Models

anthropic/claude-3-sonnet
anthropic/claude-3-opus
openai/gpt-4
deepseek/deepseek-r1

Error Responses

400: Failed to initialize AI model
401: No Bearer token provided
500: Internal server error

Usage in Agent Requests

When making requests to agent endpoints, specify the model in the request body:

{
  "prompt": "Your task description",
  "model": "anthropic/claude-3-sonnet",
  "context": "Additional context"
}

Best Practices

Match model to task: Use specialized models for their strengths
Consider context length: Choose models with sufficient context for your data
Monitor costs: Balance performance with cost considerations
Test different models: Experiment to find the best model for your use case
Use free tier models: Start with free models for development and testing

Agents Authentication