Anthropic - Docs | Transactional

Overview

AI Gateway supports Anthropic's Claude models with automatic translation from the OpenAI API format. Use the familiar OpenAI SDK to access Claude Opus 4, Claude Sonnet 4, Claude 3.5, and Claude 3 Haiku.

Supported Models

Model	Context Window	Max Output	Best For
`claude-opus-4`	200K	32K	Most capable, complex tasks
`claude-sonnet-4`	200K	64K	Balanced performance
`claude-3-5-sonnet`	200K	8K	General purpose, fast
`claude-3-opus`	200K	4K	Previous flagship
`claude-3-sonnet`	200K	4K	Previous generation
`claude-3-haiku`	200K	4K	Fast, cost-effective

Setup

1. Get Anthropic API Key

Go to Anthropic Console
Create a new API key
Copy the key (starts with sk-ant-)

2. Add Key to AI Gateway

Navigate to AI Gateway Settings
Click Add Key under Provider Keys
Select "Anthropic"
Paste your API key
Click Save

Usage

Basic Chat Completion

Use the OpenAI SDK format - AI Gateway handles the translation:

import OpenAI from 'openai';
 
const openai = new OpenAI({
  baseURL: 'https://api.transactional.dev/ai/v1',
  apiKey: process.env.GATEWAY_API_KEY,
});
 
const response = await openai.chat.completions.create({
  model: 'claude-3-5-sonnet',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'What is the capital of France?' }
  ],
});
 
console.log(response.choices[0].message.content);

With Streaming

const stream = await openai.chat.completions.create({
  model: 'claude-3-5-sonnet',
  messages: [{ role: 'user', content: 'Tell me a story' }],
  stream: true,
});
 
for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

With Tool Use

Claude's tool use is supported via OpenAI's function calling format:

const response = await openai.chat.completions.create({
  model: 'claude-3-5-sonnet',
  messages: [{ role: 'user', content: 'What is the weather in Paris?' }],
  tools: [
    {
      type: 'function',
      function: {
        name: 'get_weather',
        description: 'Get the current weather for a location',
        parameters: {
          type: 'object',
          properties: {
            location: { type: 'string', description: 'City name' }
          },
          required: ['location']
        }
      }
    }
  ],
});

Long Context

Claude supports up to 200K tokens of context:

const response = await openai.chat.completions.create({
  model: 'claude-3-5-sonnet',
  messages: [
    { role: 'user', content: veryLongDocument }, // Up to 200K tokens
    { role: 'user', content: 'Summarize the key points' }
  ],
});

OpenAI to Anthropic Translation

AI Gateway automatically handles format differences:

OpenAI Format	Anthropic Format
`system` message	`system` parameter
`function`	`tool_use`
`tool_choice: "auto"`	`tool_choice: { type: "auto" }`
`response_format: json`	JSON instruction in prompt

System Messages

System messages are automatically extracted and sent as Anthropic's system parameter:

// Your code (OpenAI format)
messages: [
  { role: 'system', content: 'You are helpful.' },
  { role: 'user', content: 'Hello' }
]
 
// AI Gateway sends to Anthropic
{
  system: 'You are helpful.',
  messages: [
    { role: 'user', content: 'Hello' }
  ]
}

Parameters

Supported parameters with automatic translation:

OpenAI Parameter	Anthropic Equivalent
`temperature`	`temperature` (0-1, scaled)
`max_tokens`	`max_tokens`
`top_p`	`top_p`
`stop`	`stop_sequences`
`stream`	`stream`
`tools`	`tools`

Parameter Differences

Temperature: OpenAI uses 0-2, Anthropic uses 0-1. AI Gateway scales automatically.
Max Tokens: Required for Anthropic. AI Gateway sets a default if not provided.

Pricing

AI Gateway passes through Anthropic's pricing. Current rates (as of Jan 2025):

Model	Input	Output
claude-opus-4	$15.00/1M	$75.00/1M
claude-sonnet-4	$3.00/1M	$15.00/1M
claude-3-5-sonnet	$3.00/1M	$15.00/1M
claude-3-opus	$15.00/1M	$75.00/1M
claude-3-haiku	$0.25/1M	$1.25/1M

Best Practices

Model Selection

claude-opus-4: Complex reasoning, analysis, code generation
claude-sonnet-4: Balanced performance for most tasks
claude-3-5-sonnet: Fast responses, general purpose
claude-3-haiku: Quick tasks, lower cost

System Prompts

Claude responds well to clear, detailed system prompts:

const response = await openai.chat.completions.create({
  model: 'claude-3-5-sonnet',
  messages: [
    {
      role: 'system',
      content: `You are an expert technical writer.
        - Use clear, concise language
        - Include code examples when relevant
        - Format responses in markdown`
    },
    { role: 'user', content: 'Explain async/await in JavaScript' }
  ],
});

Troubleshooting

Model not found

Ensure you've added your Anthropic API key in Settings.

Rate limits

Anthropic has different rate limits than OpenAI. Check your Anthropic console.

Long responses cut off

Always set max_tokens for Claude models:

const response = await openai.chat.completions.create({
  model: 'claude-3-5-sonnet',
  max_tokens: 4096,  // Always set for Anthropic
  messages: [...],
});

Next Steps

OpenAI Integration
Fallback Configuration - Use Claude as fallback
Cost Tracking - Monitor Claude costs