Supported Providers

Supported Providers

UniCraft supports a wide range of AI providers, allowing you to access multiple models through a single, unified interface.

Provider Overview

Provider	Status	Models	Pricing	Features
OpenAI	✅ Active	GPT-4, GPT-3.5, Embeddings	Pay-per-token	Chat, Completions, Embeddings
Anthropic	✅ Active	Claude 3 Haiku, Sonnet, Opus	Pay-per-token	Chat, Completions
Google	✅ Active	Gemini Pro, PaLM 2	Pay-per-token	Chat, Completions, Embeddings
Cohere	✅ Active	Command, Embed	Pay-per-token	Chat, Completions, Embeddings
Hugging Face	✅ Active	Various open models	Pay-per-token	Chat, Completions, Embeddings

OpenAI

Overview

OpenAI is one of the leading AI providers, offering state-of-the-art language models.

Supported Models

Chat Models

• gpt-4: Most capable model for complex tasks
• gpt-4-turbo: Faster version of GPT-4
• gpt-3.5-turbo: Cost-effective model for most tasks
• gpt-3.5-turbo-16k: Extended context version

Embedding Models

• text-embedding-ada-002: General-purpose embeddings
• text-embedding-3-small: Smaller, faster embeddings
• text-embedding-3-large: Larger, more accurate embeddings

Configuration

const openaiProvider = {
  name: "OpenAI Production",
  type: "openai",
  api_key: process.env.OPENAI_API_KEY,
  models: ["gpt-4", "gpt-3.5-turbo"],
  priority: 1,
  enabled: true,
};

Pricing (as of 2024)

• GPT-4: $0.03/1K input tokens, $0.06/1K output tokens
• GPT-3.5-turbo: $0.001/1K input tokens, $0.002/1K output tokens
• Embeddings: $0.0001/1K tokens

Anthropic

Overview

Anthropic provides Claude models, known for their helpfulness and safety.

Supported Models

Chat Models

• claude-3-opus: Most capable model for complex tasks
• claude-3-sonnet: Balanced performance and cost
• claude-3-haiku: Fast and cost-effective

Configuration

const anthropicProvider = {
  name: "Anthropic Production",
  type: "anthropic",
  api_key: process.env.ANTHROPIC_API_KEY,
  models: ["claude-3-sonnet", "claude-3-haiku"],
  priority: 2,
  enabled: true,
};

Pricing (as of 2024)

• Claude 3 Opus: $0.015/1K input tokens, $0.075/1K output tokens
• Claude 3 Sonnet: $0.003/1K input tokens, $0.015/1K output tokens
• Claude 3 Haiku: $0.00025/1K input tokens, $0.00125/1K output tokens

Google

Overview

Google provides Gemini and PaLM models through their AI Studio.

Supported Models

Chat Models

• gemini-pro: General-purpose model
• gemini-pro-vision: Multimodal model with vision capabilities
• palm-2: Previous generation model

Configuration

const googleProvider = {
  name: "Google Production",
  type: "google",
  api_key: process.env.GOOGLE_API_KEY,
  models: ["gemini-pro"],
  priority: 3,
  enabled: true,
};

Pricing (as of 2024)

• Gemini Pro: $0.0005/1K input tokens, $0.0015/1K output tokens
• PaLM 2: $0.0005/1K input tokens, $0.0015/1K output tokens

Cohere

Overview

Cohere provides Command and Embed models for various use cases.

Supported Models

Chat Models

• command: General-purpose chat model
• command-light: Lighter version for faster responses

Embedding Models

• embed-english-v2.0: English embeddings
• embed-multilingual-v2.0: Multilingual embeddings

Configuration

const cohereProvider = {
  name: "Cohere Production",
  type: "cohere",
  api_key: process.env.COHERE_API_KEY,
  models: ["command", "command-light"],
  priority: 4,
  enabled: true,
};

Pricing (as of 2024)

• Command: $0.001/1K input tokens, $0.002/1K output tokens
• Embeddings: $0.0001/1K tokens

Hugging Face

Overview

Hugging Face provides access to various open-source models.

Supported Models

Chat Models

• meta-llama/Llama-2-70b-chat-hf: Llama 2 70B
• mistralai/Mistral-7B-Instruct-v0.1: Mistral 7B
• microsoft/DialoGPT-medium: DialoGPT

Embedding Models

• sentence-transformers/all-MiniLM-L6-v2: General-purpose embeddings
• sentence-transformers/all-mpnet-base-v2: High-quality embeddings

Configuration

const huggingfaceProvider = {
  name: "Hugging Face Production",
  type: "huggingface",
  api_key: process.env.HUGGINGFACE_API_KEY,
  models: ["meta-llama/Llama-2-70b-chat-hf"],
  priority: 5,
  enabled: true,
};

Pricing (as of 2024)

• Varies by model and usage
• Pay-per-token pricing
• Some models available for free

Custom Providers

Overview

UniCraft supports custom providers for specialized use cases.

Configuration

const customProvider = {
  name: "Custom Provider",
  type: "custom",
  base_url: "https://your-api.com/v1",
  api_key: process.env.CUSTOM_API_KEY,
  models: ["custom-model"],
  priority: 6,
  enabled: true,
  headers: {
    Authorization: "Bearer your-token",
    "Custom-Header": "value",
  },
};

Provider Management

Adding Providers

Via Dashboard

1. Navigate to Settings → Providers
2. Click Add Provider
3. Select provider type
4. Enter API credentials
5. Configure models and settings

Via API

const provider = await fetch("https://api.unicraft.com/v1/providers", {
  method: "POST",
  headers: {
    Authorization: "Bearer your-unicraft-api-key",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    name: "OpenAI Production",
    type: "openai",
    api_key: process.env.OPENAI_API_KEY,
    models: ["gpt-4", "gpt-3.5-turbo"],
    priority: 1,
    enabled: true,
  }),
});

Updating Providers

const updatedProvider = await fetch(
  "https://api.unicraft.com/v1/providers/provider_123",
  {
    method: "PUT",
    headers: {
      Authorization: "Bearer your-unicraft-api-key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      priority: 2,
      enabled: false,
    }),
  }
);

Testing Providers

const testResult = await fetch(
  "https://api.unicraft.com/v1/providers/provider_123/test",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer your-unicraft-api-key",
    },
  }
);

const result = await testResult.json();
console.log("Test result:", result);

Provider Status

Health Checks

UniCraft automatically monitors provider health:

• Available: Provider is responding normally
• Degraded: Provider is responding but with issues
• Unavailable: Provider is not responding
• Rate Limited: Provider is rate limiting requests

Status Monitoring

const status = await fetch("https://api.unicraft.com/v1/providers/status", {
  headers: {
    Authorization: "Bearer your-unicraft-api-key",
  },
});

const providerStatus = await status.json();
console.log("Provider status:", providerStatus);

Failover and Redundancy

Automatic Failover

Configure automatic failover when providers are unavailable:

const failoverConfig = {
  enabled: true,
  primary_provider: "openai",
  fallback_providers: ["anthropic", "google"],
  retry_attempts: 3,
  retry_delay: 1000,
};

Load Balancing

Distribute requests across multiple providers:

const loadBalancer = {
  strategy: "round_robin", // or "weighted", "least_connections"
  weights: {
    openai: 0.5,
    anthropic: 0.3,
    google: 0.2,
  },
};

Best Practices

1. Provider Selection

• Use multiple providers for redundancy
• Choose providers based on your use case
• Consider cost vs. quality trade-offs

2. Configuration

• Set appropriate priorities
• Configure rate limits
• Enable health monitoring

3. Monitoring

• Monitor provider status regularly
• Set up alerts for provider issues
• Track usage and costs

4. Security

• Store API keys securely
• Use environment variables
• Implement proper access controls

Troubleshooting

Common Issues

1. Provider Not Available

   • Check API key validity
   • Verify provider status
   • Test connection

2. Rate Limiting

   • Implement rate limiting
   • Use multiple providers
   • Monitor usage

3. High Costs

   • Optimize model selection
   • Use cost-effective providers
   • Monitor spending

4. Poor Performance

   • Check provider health
   • Optimize requests
   • Use appropriate models

Debug Tips

1. Test Connections: Test each provider individually
2. Monitor Logs: Check provider logs for issues
3. Use Health Checks: Regular health check monitoring
4. Check Status: Monitor provider status dashboard

Support

For provider-related support:

• Documentation: UniCraft Docs
• Provider Status: UniCraft Status Page
• Community: UniCraft Community
• Support: support@unicraft.com

Next Steps

After setting up providers:

1. Test All Providers: Test each provider individually
2. Configure Failover: Set up automatic failover
3. Monitor Performance: Set up monitoring and alerts
4. Optimize Costs: Monitor and optimize spending
5. Scale as Needed: Add more providers as needed