Skip to main content
OpenRouter provides unified access to models from multiple providers including OpenAI, Anthropic, Google, Meta, and more through a single API.

​
Why Use OpenRouter?

  • Single API Key: Access models from multiple providers
  • Automatic Fallback: Configure fallback models if primary is unavailable
  • Model Comparison: Test different models without multiple integrations
  • Cost Optimization: Choose the best model for your budget
  • No Vendor Lock-in: Switch between providers seamlessly

​
Setup

​
1. Get API Key

  1. Visit OpenRouter
  2. Sign up or log in
  3. Navigate to API Keys
  4. Create a new API key
  5. Copy the key (starts with sk-or-v1-)

​
2. Configure Weaver

Option A: Configuration File Add to ~/.weaver/config.json: 
{
  "providers": {
    "openrouter": {
      "api_key": "sk-or-v1-...",
      "api_base": "https://openrouter.ai/api/v1"
    }
  },
  "agents": {
    "defaults": {
      "provider": "openrouter",
      "model": "anthropic/claude-3.5-sonnet"
    }
  }
}
Option B: Environment Variable Add to .env: 
OPENROUTER_API_KEY=sk-or-v1-...

​
Model Selection

OpenRouter uses provider prefixes in model names:
Model IDProviderDescription
anthropic/claude-3.5-sonnetAnthropicLatest Claude Sonnet
openai/gpt-4-turboOpenAIGPT-4 Turbo
google/gemini-pro-1.5GoogleGemini Pro
meta-llama/llama-3.1-70b-instructMetaLlama 3.1 70B
deepseek/deepseek-chatDeepSeekDeepSeek V3

​
Model Discovery

Find available models: 
curl https://openrouter.ai/api/v1/models \
  -H "Authorization: Bearer $OPENROUTER_API_KEY"
Or browse at OpenRouter Models.

​
Configuration Options

​
api_key
string
OpenRouter API key (required)
​
api_base
string
default:"https://openrouter.ai/api/v1"
OpenRouter API endpoint
​
proxy
string
HTTP/HTTPS proxy URL for API requests (optional)

​
Automatic Provider Detection

Weaver automatically uses OpenRouter when model names include:
  • openrouter/ prefix
  • anthropic/ prefix
  • openai/ prefix
  • meta-llama/ prefix
  • deepseek/ prefix
  • google/ prefix
Source: pkg/providers/http_provider.go:372-380

​
Example

{
  "providers": {
    "openrouter": {
      "api_key": "sk-or-v1-..."
    }
  },
  "agents": {
    "defaults": {
      "model": "anthropic/claude-3.5-sonnet"
    }
  }
}
Weaver automatically uses OpenRouter because model starts with anthropic/.

​
Fallback Behavior

If no specific provider is configured, Weaver falls back to OpenRouter: 
// From http_provider.go:456-468
default:
  if cfg.Providers.OpenRouter.APIKey != "" {
    apiKey = cfg.Providers.OpenRouter.APIKey
    proxy = cfg.Providers.OpenRouter.Proxy
    if cfg.Providers.OpenRouter.APIBase != "" {
      apiBase = cfg.Providers.OpenRouter.APIBase
    } else {
      apiBase = "https://openrouter.ai/api/v1"
    }
  }
This makes OpenRouter an excellent default provider.

​
Model Parameters

Configure model behavior: 
{
  "agents": {
    "defaults": {
      "model": "anthropic/claude-3.5-sonnet",
      "max_tokens": 8192,
      "temperature": 0.7
    }
  }
}
​
max_tokens
integer
default:"8192"
Maximum tokens in response
​
temperature
float
default:"0.7"
Controls randomness (0.0 = deterministic, 2.0 = very random)

​
Usage Examples

​
Using Different Providers

# Use Claude via OpenRouter
weaver chat --model anthropic/claude-3.5-sonnet

# Use GPT-4 via OpenRouter
weaver chat --model openai/gpt-4-turbo

# Use Gemini via OpenRouter
weaver chat --model google/gemini-pro-1.5

# Use Llama via OpenRouter
weaver chat --model meta-llama/llama-3.1-70b-instruct

​
Cost Optimization

Switch between models based on task complexity: 
# Use cheaper model for simple tasks
weaver chat --model anthropic/claude-3-haiku

# Use powerful model for complex tasks
weaver chat --model anthropic/claude-3-opus

​
Model Comparison

Test the same prompt with different models: 
for model in "openai/gpt-4" "anthropic/claude-3.5-sonnet" "google/gemini-pro"; do
  echo "Testing $model"
  weaver chat --model "$model" "Explain quantum computing"
done

​
Implementation Details

Weaver uses the HTTPProvider for OpenRouter:
  • OpenAI-compatible API format
  • Standard /chat/completions endpoint
  • Automatic model namespace handling
  • Tool calling support
Source: pkg/providers/http_provider.go

​
Tool Calling

OpenRouter supports tool calling for compatible models: 
{
  "model": "anthropic/claude-3.5-sonnet",
  "messages": [...],
  "tools": [...],
  "tool_choice": "auto"
}
Source: pkg/providers/http_provider.go:69-72

​
Pricing

OpenRouter uses per-token pricing that varies by model:
  • View pricing at OpenRouter Pricing
  • Pricing shown in USD per 1M tokens
  • Some models offer free tiers
  • Billing is usage-based

​
Check Costs

curl https://openrouter.ai/api/v1/models \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  | jq '.data[] | {id: .id, pricing: .pricing}'

​
Troubleshooting

​
API Key Issues

# Verify API key
echo $OPENROUTER_API_KEY

# Test with curl
curl https://openrouter.ai/api/v1/models \
  -H "Authorization: Bearer $OPENROUTER_API_KEY"

​
Common Errors

  • Verify API key is correct and starts with sk-or-v1-
  • Check that the API key hasn’t been revoked
  • Ensure you have credits in your OpenRouter account
  • Verify model ID is correct (include provider prefix)
  • Check available models
  • Some models require allowlist access
  • OpenRouter has rate limits per model
  • Implement exponential backoff
  • Consider distributing load across multiple models
  • Add credits to your OpenRouter account
  • Check your balance at OpenRouter Dashboard
  • Some models are free for testing

​
Advanced Features

​
Model Fallback

OpenRouter supports automatic fallback: 
{
  "model": "anthropic/claude-3.5-sonnet",
  "route": "fallback",
  "models": [
    "anthropic/claude-3.5-sonnet",
    "anthropic/claude-3-haiku",
    "openai/gpt-4-turbo"
  ]
}

​
Response Metadata

OpenRouter returns usage metadata: 
{
  "usage": {
    "prompt_tokens": 100,
    "completion_tokens": 50,
    "total_tokens": 150
  }
}
Weaver automatically parses this into UsageInfo. Source: pkg/providers/http_provider.go:150

​
Next Steps

Provider Overview

Back to all providers

Model Comparison

Choose the right model

Build docs developers (and LLMs) love

Get started for freeTalk to us