Skip to content

Bring Your Own LLM

Connect your preferred AI provider directly to Lingo.dev for complete translation control.

Benefits at a Glance

  • Use Your Own AI Account: Connect OpenAI or Anthropic without a Lingo.dev account
  • Complete Model Control: Select specific models that match your needs
  • Custom Translation Prompts: Craft prompts for your exact translation style
  • Independent Management: Control your own API usage and costs
  • No Vendor Lock-in: Use your existing API keys and accounts

Quick Setup

Add your provider configuration to i18n.json:

json
{
  "version": 1.5,
  "provider": {
    "id": "anthropic",
    "model": "claude-3-7-sonnet-latest",
    "prompt": "You're translating text from {source} to {target}."
  },
  "locale": {
    "source": "en",
    "targets": ["es", "fr", "it", "de", "ar"]
  }
}

Note: The {source} and {target} tokens are automatically replaced with actual locale codes during translation.

Supported Providers

Lingo.dev (Managed Service)

json
"provider": {
  "id": "lingo",
  "model": "best",
  "prompt": ""
}

OpenAI

json
"provider": {
  "id": "openai",
  "model": "gpt-4",
  "prompt": "Translate accurately from {source} to {target}, maintaining the original meaning.",
  "baseUrl": "https://api.openai.com/v1" // Optional
}

Anthropic

json
"provider": {
  "id": "anthropic",
  "model": "claude-3-opus-20240229",
  "prompt": "You're an expert translator from {source} to {target}. Maintain the tone and style."
}

Setting API Keys

Set the appropriate environment variable for your chosen provider:

OpenAI

bash
export OPENAI_API_KEY=your_openai_api_key

Anthropic

bash
export ANTHROPIC_API_KEY=your_anthropic_api_key

Feature Comparison

FeatureLingo.dev ManagedYour Own LLM
Translation
Translation Memory❌ (requires manual tracking)
Glossary Support❌ (requires manual management)
Context-aware Translation✅ (industry + product optimized)✅ (via files + system prompt)
Custom PromptsLimited (with brand voice)✅ (complete control)
Model SelectionLimited (auto-routed to best model)✅ (any supported model)
Cost ControlManaged pricingSelf-managed (pay your provider)
Quality Assurance✅ (automated)❌ (manual verification needed)
Format Preservation✅ (guaranteed)Partial (depends on prompt)
Consistency✅ (automatic)❌ (manual effort required)
Error Handling✅ (robust recovery + Enterprise SLA)Basic retries only

When to Choose Each Option

Best for Lingo.dev Managed Service:

  • Projects requiring consistent translations across multiple files
  • When translation memory and glossary features are important
  • Enterprise projects with strict quality requirements
  • Teams needing centralized translation management
  • When you want a worry-free, quality-assured solution

Best for Your Own LLM:

  • Projects needing highly customized translation styles
  • When you require specific AI models
  • If you already have API credits with OpenAI or Anthropic
  • Experimental or specialized translation projects
  • When you have prompt engineering expertise to leverage

Using Your Own LLM

  1. Configure your provider in i18n.json
  2. Set the appropriate API key as an environment variable
  3. Run translations with the standard command:
bash
# Example workflow
export ANTHROPIC_API_KEY=your_api_key
npx lingo.dev@latest i18n

This approach gives you provider flexibility while still benefiting from Lingo.dev's workflow tools.

Request New Provider Support

Want support for additional AI providers? We welcome your input:

  • Open an issue on our GitHub repository
  • Submit a pull request with your provider implementation

We continuously expand our provider options based on community feedback.