Bring Your Own Key (BYOK) Setup#

Configure direct provider access using your own API keys for enhanced performance and cost optimization.

Overview#

BYOK allows you to use your existing OpenAI, Anthropic, and Groq API keys directly through WatchLLM. This provides:

• Better Performance: Direct provider access with lower latency
• Cost Optimization: Bypass intermediary pricing
• Full Model Access: Access to all provider models and features
• Enhanced Privacy: Your keys never leave your infrastructure

Supported Providers#

Setup Instructions#

1. Obtain API Keys#

OpenAI:

1. Go to platform.openai.com/api-keys
2. Create a new API key
3. Copy the key (starts with sk-)

Anthropic:

1. Go to console.anthropic.com
2. Navigate to API Keys
3. Create a new key
4. Copy the key (starts with sk-ant-)

Groq:

1. Go to console.groq.com/keys
2. Create a new API key
3. Copy the key (starts with gsk_)

2. Configure in WatchLLM#

1. Login to your WatchLLM dashboard
2. Navigate to Project Settings → Provider Keys
3. Add Keys for each provider you want to use:

   OpenAI API Key: sk-...
   Anthropic API Key: sk-ant-...
   Groq API Key: gsk_...
   

4. Save Changes

3. Update Your Code#

Once BYOK is configured, use native model names directly:

import OpenAI from 'openai';
 
const client = new OpenAI({
  apiKey: 'lgw_proj_your_watchllm_key',
  baseURL: 'https://proxy.watchllm.dev/v1'
});
 
// Now works with native provider models
const response = await client.chat.completions.create({
  model: 'gpt-4o', // Direct OpenAI
  messages: [{ role: 'user', content: 'Hello!' }]
});
 
// Anthropic models
const claudeResponse = await client.chat.completions.create({
  model: 'claude-3-sonnet-20240229', // Direct Anthropic
  messages: [{ role: 'user', content: 'Hello!' }]
});

Model Name Mapping#

Security & Encryption#

• End-to-End Encryption: Keys are encrypted at rest using AES-256
• Zero-Knowledge: WatchLLM cannot access your provider accounts
• Scoped Access: Keys are project-specific and can be rotated anytime
• Audit Logs: All key usage is logged for security monitoring

Billing & Costs#

• Direct Billing: You pay providers directly for API usage
• WatchLLM Fees: Only pay for WatchLLM's caching infrastructure
• Cost Transparency: See exactly what each request costs via response headers

Troubleshooting#

"Model not found" Error#

• Ensure you've added the correct API key for that provider
• Check that the model name matches the BYOK format (not OpenRouter format)

Authentication Failed#

• Verify your API keys are valid and have sufficient credits
• Check that keys are entered correctly in project settings
• Try rotating the key if issues persist

Still Using OpenRouter#

• BYOK takes precedence over OpenRouter for configured providers
• Clear your browser cache if model selection seems stuck

Migration from OpenRouter#

If you're currently using OpenRouter models, here's how to migrate:

// Before (OpenRouter)
const response = await client.chat.completions.create({
  model: 'openai/gpt-4o',
  messages: [{ role: 'user', content: 'Hello!' }]
});
 
// After (BYOK)
const response = await client.chat.completions.create({
  model: 'gpt-4o', // Direct OpenAI access
  messages: [{ role: 'user', content: 'Hello!' }]
});

The caching behavior and response format remain identical.