Integration Guide — Updated June 2026
How to Use Sakana Fugu: Complete API Setup Guide
Everything you need to integrate Sakana Fugu into your workflow — from account setup to API calls. Sakana Fugu works with any OpenAI-compatible client, so you can start in minutes.
Quick Start: Sakana Fugu in 4 Steps
You Send a Request
Call the Sakana Fugu API exactly like you'd call OpenAI. One endpoint, standard format. No SDK changes needed — if your code works with GPT, it works with Sakana Fugu.
Fugu Analyzes the Task
Sakana Fugu's conductor model (built on ICLR 2026 research) analyzes your request and determines which expert models to activate. It assigns Thinker, Worker, and Verifier roles dynamically.
Agents Coordinate
Multiple frontier models work together behind the scenes. Sakana Fugu orchestrates their communication using learned natural-language coordination strategies — not hand-designed workflows.
Unified Response
Sakana Fugu synthesizes all agent outputs into a single, coherent response. You receive one answer — the best result the coordinated system could produce.
Sakana Fugu API: Python Example
Sakana Fugu uses the standard OpenAI SDK format. If you already use the openai Python package, switching to Sakana Fugu requires only changing the base_url and api_key:
from openai import OpenAI
# Initialize the Sakana Fugu client
client = OpenAI(
base_url="https://api.sakana.ai/v1",
api_key="your-sakana-fugu-api-key"
)
# Call Sakana Fugu (balanced)
response = client.chat.completions.create(
model="fugu",
messages=[
{"role": "user", "content": "Review this code for bugs"}
]
)
print(response.choices[0].message.content)
# Or use Sakana Fugu Ultra for complex tasks
response = client.chat.completions.create(
model="fugu-ultra-20260615",
messages=[
{"role": "user", "content": "Analyze this research paper"}
]
)Sakana Fugu API: curl Example
You can also call Sakana Fugu directly via curl using standard OpenAI chat completion format:
curl https://api.sakana.ai/v1/chat/completions \
-H "Authorization: Bearer your-sakana-fugu-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "fugu",
"messages": [
{"role": "user", "content": "Explain quantum computing"}
],
"stream": true
}'Sakana Fugu Model Selection Guide
Choose the right Sakana Fugu model for your task:
| Model | Model ID | Best For | Latency |
|---|---|---|---|
| Sakana Fugu | fugu | Coding, chat, code review, daily tasks | Low |
| Sakana Fugu Ultra | fugu-ultra-20260615 | Research, competitions, security, complex reasoning | Higher |
Tips for Getting the Most from Sakana Fugu
Use Fugu Ultra for Multi-Step Tasks
When your task involves multiple stages — research, analysis, synthesis — Sakana Fugu Ultra's deeper agent pool will significantly outperform the balanced model. The extra latency pays for itself in answer quality.
Leverage Streaming for Interactive Use
Enable streaming (stream: true) when using Sakana Fugu in chatbots or interactive applications. Sakana Fugu streams the final response as it is synthesized, keeping the user experience responsive.
Monitor Token Usage via Response Headers
Sakana Fugu provides real-time token usage and cost tracking in API response headers. Use this data to optimize your Sakana Fugu spending and choose between subscription and pay-as-you-go billing.
Opt Out Agents for Compliance
If your organization has data residency or vendor restrictions, use Sakana Fugu's agent opt-out feature to exclude specific models from the orchestration pool. Configure this in your Sakana Fugu console.
Start with Fugu, Escalate to Ultra
For cost efficiency, route most requests through Sakana Fugu (balanced) and escalate only complex or high-stakes tasks to Sakana Fugu Ultra. This tiered approach maximizes the value of your Sakana Fugu subscription.