How to Use Fugu AI’s API: A Step-by-Step Guide for Developers (2026)

Getting Started with Fugu AI’s API in 2026

Fugu AI’s orchestration system lets you delegate sub-tasks to multiple frontier models through a single API endpoint. This guide walks you through the setup, from basic calls to advanced research workflows using Fugu Ultra.

Prerequisites

A Fugu AI account (sign up at fugu.ai)
API key from the dashboard
Basic knowledge of REST APIs and JSON

Step 1: Authentication

Include your API key in the request header. Example using cURL:

curl -X POST https://api.fugu.ai/v1/chat 
  -H 'Authorization: Bearer YOUR_API_KEY' 
  -H 'Content-Type: application/json'

Step 2: Making a Standard Request

Send a simple task to Fugu’s default model. The API accepts a messages array similar to OpenAI’s format.

{
  "model": "fugu-default",
  "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Explain quantum computing in simple terms."}
  ]
}

Response includes the model’s reply and metadata about which sub-models were used.

Step 3: Delegating Sub-Tasks

Fugu’s orchestration automatically breaks complex tasks into sub-tasks and assigns them to the best frontier models. You can specify preferences using the orchestration parameter.

{
  "model": "fugu-default",
  "messages": [{"role": "user", "content": "Write a Python script to scrape a website and summarize the content."}],
  "orchestration": {
    "prefer_speed": true,
    "max_subtasks": 5
  }
}

Available Orchestration Options

prefer_speed: Prioritizes faster models for sub-tasks
prefer_accuracy: Uses more capable models for critical sub-tasks
max_subtasks: Limits the number of sub-tasks created

Step 4: Using Fugu Ultra for Complex Research

Fugu Ultra is designed for deep research tasks. It uses multiple reasoning steps and can access external tools.

{
  "model": "fugu-ultra",
  "messages": [{"role": "user", "content": "Analyze the latest trends in renewable energy for 2026."}],
  "ultra_config": {
    "research_depth": "deep",
    "include_citations": true
  }
}

Response includes a structured report with citations and sub-task breakdowns.

Step 5: Handling Responses

Parse the JSON response to extract the main content and metadata.

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "created": 1700000000,
  "model": "fugu-default",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "Here is the Python script..."
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 50,
    "completion_tokens": 200,
    "total_tokens": 250
  },
  "orchestration_details": {
    "subtasks": [
      {"task": "code_generation", "model": "gpt-4"},
      {"task": "summarization", "model": "claude-3"}
    ]
  }
}

Best Practices

Start with simple tasks to test orchestration behavior
Use prefer_accuracy for tasks requiring high reliability
Monitor usage via the dashboard to optimize costs
Cache responses for repeated queries

Troubleshooting Common Issues

Rate Limiting

If you receive 429 errors, implement exponential backoff. Free tier allows 100 requests per minute.

Sub-Task Failures

Fugu automatically retries failed sub-tasks. Check orchestration_details for retry counts.

Conclusion

Fugu AI’s API simplifies complex AI workflows by handling model orchestration behind the scenes. Start with standard requests, then explore Fugu Ultra for deep research. The key is to let Fugu decide which models to use for each sub-task, saving you development time.

Connect