Skip to content
UniCraft AI Model Router

Supported Models

Supported Models

Section titled “Supported Models”

UniCraft supports a wide range of AI models from multiple providers, allowing you to choose the best model for your specific use case.

Model Categories

Section titled “Model Categories”

Chat Models

Section titled “Chat Models”

Models designed for conversational AI and chat applications.

Completion Models

Section titled “Completion Models”

Models for text completion and generation tasks.

Embedding Models

Section titled “Embedding Models”

Models for converting text into vector representations.

Multimodal Models

Section titled “Multimodal Models”

Models that can process both text and images.

OpenAI Models

Section titled “OpenAI Models”

GPT-4 Series

Section titled “GPT-4 Series”

GPT-4

Section titled “GPT-4”
  • Model ID: gpt-4
  • Context Length: 8,192 tokens
  • Best For: Complex reasoning, analysis, creative writing
  • Pricing: $0.03/1K input, $0.06/1K output
  • Availability: High

GPT-4 Turbo

Section titled “GPT-4 Turbo”
  • Model ID: gpt-4-turbo
  • Context Length: 128,000 tokens
  • Best For: Large context tasks, document analysis
  • Pricing: $0.01/1K input, $0.03/1K output
  • Availability: High

GPT-4 Vision

Section titled “GPT-4 Vision”
  • Model ID: gpt-4-vision-preview
  • Context Length: 128,000 tokens
  • Best For: Image analysis, multimodal tasks
  • Pricing: $0.01/1K input, $0.03/1K output
  • Availability: Medium

GPT-3.5 Series

Section titled “GPT-3.5 Series”

GPT-3.5 Turbo

Section titled “GPT-3.5 Turbo”
  • Model ID: gpt-3.5-turbo
  • Context Length: 4,096 tokens
  • Best For: General-purpose tasks, cost-effective
  • Pricing: $0.001/1K input, $0.002/1K output
  • Availability: High

GPT-3.5 Turbo 16K

Section titled “GPT-3.5 Turbo 16K”
  • Model ID: gpt-3.5-turbo-16k
  • Context Length: 16,384 tokens
  • Best For: Longer conversations, document processing
  • Pricing: $0.003/1K input, $0.004/1K output
  • Availability: High

Embedding Models

Section titled “Embedding Models”

Text Embedding Ada 002

Section titled “Text Embedding Ada 002”
  • Model ID: text-embedding-ada-002
  • Dimensions: 1,536
  • Best For: General-purpose embeddings
  • Pricing: $0.0001/1K tokens
  • Availability: High

Text Embedding 3 Small

Section titled “Text Embedding 3 Small”
  • Model ID: text-embedding-3-small
  • Dimensions: 1,536
  • Best For: Faster, cost-effective embeddings
  • Pricing: $0.00002/1K tokens
  • Availability: High

Text Embedding 3 Large

Section titled “Text Embedding 3 Large”
  • Model ID: text-embedding-3-large
  • Dimensions: 3,072
  • Best For: High-quality embeddings
  • Pricing: $0.00013/1K tokens
  • Availability: High

Anthropic Models

Section titled “Anthropic Models”

Claude 3 Series

Section titled “Claude 3 Series”

Claude 3 Opus

Section titled “Claude 3 Opus”
  • Model ID: claude-3-opus-20240229
  • Context Length: 200,000 tokens
  • Best For: Complex reasoning, analysis, creative tasks
  • Pricing: $0.015/1K input, $0.075/1K output
  • Availability: High

Claude 3 Sonnet

Section titled “Claude 3 Sonnet”
  • Model ID: claude-3-sonnet-20240229
  • Context Length: 200,000 tokens
  • Best For: Balanced performance and cost
  • Pricing: $0.003/1K input, $0.015/1K output
  • Availability: High

Claude 3 Haiku

Section titled “Claude 3 Haiku”
  • Model ID: claude-3-haiku-20240307
  • Context Length: 200,000 tokens
  • Best For: Fast, cost-effective responses
  • Pricing: $0.00025/1K input, $0.00125/1K output
  • Availability: High

Google Models

Section titled “Google Models”

Gemini Series

Section titled “Gemini Series”

Gemini Pro

Section titled “Gemini Pro”
  • Model ID: gemini-pro
  • Context Length: 32,000 tokens
  • Best For: General-purpose tasks, multilingual
  • Pricing: $0.0005/1K input, $0.0015/1K output
  • Availability: High

Gemini Pro Vision

Section titled “Gemini Pro Vision”
  • Model ID: gemini-pro-vision
  • Context Length: 16,000 tokens
  • Best For: Image analysis, multimodal tasks
  • Pricing: $0.0005/1K input, $0.0015/1K output
  • Availability: Medium

PaLM 2 Series

Section titled “PaLM 2 Series”

PaLM 2 Text

Section titled “PaLM 2 Text”
  • Model ID: palm-2-text-bison
  • Context Length: 8,192 tokens
  • Best For: Text generation, summarization
  • Pricing: $0.0005/1K input, $0.0015/1K output
  • Availability: High

PaLM 2 Chat

Section titled “PaLM 2 Chat”
  • Model ID: palm-2-chat-bison
  • Context Length: 8,192 tokens
  • Best For: Conversational AI, chat applications
  • Pricing: $0.0005/1K input, $0.0015/1K output
  • Availability: High

Cohere Models

Section titled “Cohere Models”

Command Series

Section titled “Command Series”

Command

Section titled “Command”
  • Model ID: command
  • Context Length: 4,096 tokens
  • Best For: General-purpose tasks, enterprise use
  • Pricing: $0.001/1K input, $0.002/1K output
  • Availability: High

Command Light

Section titled “Command Light”
  • Model ID: command-light
  • Context Length: 4,096 tokens
  • Best For: Fast, cost-effective responses
  • Pricing: $0.0005/1K input, $0.001/1K output
  • Availability: High

Embedding Models

Section titled “Embedding Models”

Embed English

Section titled “Embed English”
  • Model ID: embed-english-v2.0
  • Dimensions: 1,024
  • Best For: English text embeddings
  • Pricing: $0.0001/1K tokens
  • Availability: High

Embed Multilingual

Section titled “Embed Multilingual”
  • Model ID: embed-multilingual-v2.0
  • Dimensions: 1,024
  • Best For: Multilingual text embeddings
  • Pricing: $0.0001/1K tokens
  • Availability: High

Hugging Face Models

Section titled “Hugging Face Models”

Llama 2 Series

Section titled “Llama 2 Series”

Llama 2 70B Chat

Section titled “Llama 2 70B Chat”
  • Model ID: meta-llama/Llama-2-70b-chat-hf
  • Context Length: 4,096 tokens
  • Best For: Open-source alternative, research
  • Pricing: Varies
  • Availability: Medium

Llama 2 13B Chat

Section titled “Llama 2 13B Chat”
  • Model ID: meta-llama/Llama-2-13b-chat-hf
  • Context Length: 4,096 tokens
  • Best For: Balanced performance and cost
  • Pricing: Varies
  • Availability: High

Llama 2 7B Chat

Section titled “Llama 2 7B Chat”
  • Model ID: meta-llama/Llama-2-7b-chat-hf
  • Context Length: 4,096 tokens
  • Best For: Fast, cost-effective responses
  • Pricing: Varies
  • Availability: High

Mistral Series

Section titled “Mistral Series”

Mistral 7B Instruct

Section titled “Mistral 7B Instruct”
  • Model ID: mistralai/Mistral-7B-Instruct-v0.1
  • Context Length: 8,192 tokens
  • Best For: Fast, efficient responses
  • Pricing: Varies
  • Availability: High

Embedding Models

Section titled “Embedding Models”

All-MiniLM-L6-v2

Section titled “All-MiniLM-L6-v2”
  • Model ID: sentence-transformers/all-MiniLM-L6-v2
  • Dimensions: 384
  • Best For: General-purpose embeddings
  • Pricing: Varies
  • Availability: High

All-mpnet-base-v2

Section titled “All-mpnet-base-v2”
  • Model ID: sentence-transformers/all-mpnet-base-v2
  • Dimensions: 768
  • Best For: High-quality embeddings
  • Pricing: Varies
  • Availability: High

Model Selection Guide

Section titled “Model Selection Guide”

By Use Case

Section titled “By Use Case”

Content Generation

Section titled “Content Generation”
  • Best: GPT-4, Claude 3 Opus
  • Good: GPT-3.5 Turbo, Claude 3 Sonnet
  • Budget: GPT-3.5 Turbo, Claude 3 Haiku

Code Generation

Section titled “Code Generation”
  • Best: GPT-4, Claude 3 Opus
  • Good: GPT-3.5 Turbo, Claude 3 Sonnet
  • Budget: GPT-3.5 Turbo, Claude 3 Haiku

Analysis and Reasoning

Section titled “Analysis and Reasoning”
  • Best: GPT-4, Claude 3 Opus
  • Good: Claude 3 Sonnet, GPT-3.5 Turbo
  • Budget: Claude 3 Haiku, GPT-3.5 Turbo

Simple Q&A

Section titled “Simple Q&A”
  • Best: GPT-3.5 Turbo, Claude 3 Haiku
  • Good: Gemini Pro, Command Light
  • Budget: GPT-3.5 Turbo, Claude 3 Haiku

Embeddings

Section titled “Embeddings”
  • Best: Text Embedding 3 Large, All-mpnet-base-v2
  • Good: Text Embedding Ada 002, All-MiniLM-L6-v2
  • Budget: Text Embedding 3 Small, All-MiniLM-L6-v2

By Cost

Section titled “By Cost”

Most Expensive

Section titled “Most Expensive”
  1. GPT-4
  2. Claude 3 Opus
  3. GPT-4 Turbo

Most Cost-Effective

Section titled “Most Cost-Effective”
  1. Claude 3 Haiku
  2. GPT-3.5 Turbo
  3. Gemini Pro

By Speed

Section titled “By Speed”

Fastest

Section titled “Fastest”
  1. Claude 3 Haiku
  2. GPT-3.5 Turbo
  3. Command Light

Most Capable

Section titled “Most Capable”
  1. GPT-4
  2. Claude 3 Opus
  3. GPT-4 Turbo

Model Configuration

Section titled “Model Configuration”

Basic Configuration

Section titled “Basic Configuration”
const modelConfig = {
  model: "gpt-3.5-turbo",
  temperature: 0.7,
  max_tokens: 1000,
  top_p: 0.9,
};

Advanced Configuration

Section titled “Advanced Configuration”
const advancedConfig = {
  model: "gpt-4",
  temperature: 0.7,
  max_tokens: 2000,
  top_p: 0.9,
  frequency_penalty: 0.0,
  presence_penalty: 0.0,
  stop: ["\n\n", "Human:", "Assistant:"],
};

Smart Routing Configuration

Section titled “Smart Routing Configuration”
const smartRouting = {
  model: "auto",
  routing_strategy: "cost_optimized",
  max_cost_per_request: 0.01,
  quality_threshold: 0.8,
  preferred_models: ["gpt-3.5-turbo", "claude-3-haiku"],
};

Model Testing

Section titled “Model Testing”

Test Different Models

Section titled “Test Different Models”
const models = ["gpt-3.5-turbo", "claude-3-haiku", "gemini-pro"];
const testPrompt = "Explain quantum computing in simple terms";


for (const model of models) {
  const response = await unicraft.chat.completions.create({
    model: model,
    messages: [{ role: "user", content: testPrompt }],
    max_tokens: 200,
  });


  console.log(`${model}: ${response.choices[0].message.content}`);
}

Compare Performance

Section titled “Compare Performance”
const performanceTest = async (model, prompt) => {
  const start = Date.now();
  const response = await unicraft.chat.completions.create({
    model: model,
    messages: [{ role: "user", content: prompt }],
    max_tokens: 100,
  });
  const end = Date.now();


  return {
    model: model,
    response_time: end - start,
    cost: response.unicraft.cost,
    quality: response.unicraft.quality_score,
  };
};

Best Practices

Section titled “Best Practices”

1. Model Selection

Section titled “1. Model Selection”
  • Choose models based on your specific use case
  • Consider cost vs. quality trade-offs
  • Test multiple models before deciding

2. Configuration

Section titled “2. Configuration”
  • Use appropriate temperature settings
  • Set reasonable max_tokens limits
  • Configure stop sequences when needed

3. Optimization

Section titled “3. Optimization”
  • Use smart routing for automatic model selection
  • Implement caching for repeated requests
  • Monitor costs and performance

4. Monitoring

Section titled “4. Monitoring”
  • Track model performance metrics
  • Monitor costs and usage
  • Set up alerts for issues

Troubleshooting

Section titled “Troubleshooting”

Common Issues

Section titled “Common Issues”

  1. Model Not Available

    • Check provider configuration
    • Verify model availability
    • Use alternative models

  2. Poor Quality Responses

    • Adjust temperature settings
    • Improve prompt quality
    • Try different models

  3. High Costs

    • Use cost-effective models
    • Optimize prompts
    • Implement caching

  4. Slow Responses

    • Use faster models
    • Optimize requests
    • Check provider status

Debug Tips

Section titled “Debug Tips”
  1. Test Models: Test different models with your use case
  2. Monitor Metrics: Track performance and cost metrics
  3. Optimize Prompts: Improve prompt quality and structure
  4. Use Smart Routing: Let UniCraft choose the best model

Support

Section titled “Support”

For model-related support:

Next Steps

Section titled “Next Steps”

After selecting models:

  1. Test Models: Test different models with your use case
  2. Configure Routing: Set up smart routing rules
  3. Monitor Performance: Track model performance and costs
  4. Optimize Usage: Optimize based on performance data
  5. Scale as Needed: Add more models as needed