API Guides

Learn how to use Yotta’s APIs to manage GPUs, deploy workloads, and automate infrastructure.

Overview

This API is your gateway to building, scaling, and automating AI workloads on Yotta’s distributed GPU cloud. With just a few API calls, you can spin up compute Pods, query available GPUs, and orchestrate high-performance training or inference workloads seamlessly across regions.

---

🌐 Base URL

All API requests are made over HTTPS.

https://api.yottalabs.ai

Yotta Platform APIs follow RESTful conventions and return JSON-encoded responses. Requests sent over insecure HTTP will be rejected.

---

🔐 Authentication

Yotta Platform uses API Key-based authentication. You can create and manage your keys from the Access Keys Settings. On Yotta Console, go to Settings -> Access Keys, Copy the API Keys.

Every request must include the following header:

x-api-key: <YOUR_API_KEY>

Replace <YOUR_API_KEY> with your real key. Keep it private — do NOT share or embed it in client-side code.

---

🧭 Global Response Format

All endpoints return a consistent JSON structure:

{
  "message": "success",
  "code": 10000,
  "data": {...}
}

Field	Type	Description
message	string	Human-readable message describing result
code	integer	Response code (10000 = success)
data	object / list / null	Response data payload

For different return types, refer to Data Models.

---

🧠 Example: Create a Pod via API

Example in curl

curl -X POST 'https://api.yottalabs.ai/openapi/v1/pods/create' \
  -H 'x-api-key: <YOUR_API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{
    "image": "yottalabsai/pytorch:2.8.0-py3.11-cuda12.8.1-cudnn-devel-ubuntu22.04-2025050802",
    "gpuType": "NVIDIA_L4_24G",
    "gpuCount": 1,
    "environmentVars": [
      { "key": "JUPYTER_PASSWORD", "value": "your_password" }
    ],
    "expose": [
      { "port": 22, "protocol": "SSH" }
    ]
  }'

Example in Python

import requests

url = "https://api.yottalabs.ai/openapi/v1/pods/create"
headers = {
    "x-api-key": "<YOUR_API_KEY>",
    "Content-Type": "application/json"
}
payload = {
    "image": "yottalabsai/pytorch:2.8.0-py3.11-cuda12.8.1-cudnn-devel-ubuntu22.04-2025050802",
    "gpuType": "NVIDIA_L4_24G",
    "gpuCount": 1,
    "environmentVars": [
        {"key": "JUPYTER_PASSWORD", "value": "your_password"}
    ],
    "expose": [
        {"port": 22, "protocol": "SSH"}
    ]
}

response = requests.post(url, headers=headers, json=payload)
print(response.json())

---

⚙️ Versioning

All endpoints currently use the /openapi/v1/ prefix. Backward-compatible changes may be added to this version. When future breaking changes occur, a new version (e.g., /openapi/v2/) will be introduced with full release notes.

---

🚦 Rate Limits & Best Practices

To ensure fair usage and system stability, Yotta applies rate-limiting policies. While limits may vary by plan, the following best practices are encouraged:

Type	Recommendation
Max Requests / Minute	60–120 (depending on plan)
Concurrent Pod Operations	Up to 5 simultaneous per API key
Retry Strategy	Use exponential backoff (e.g., 1s → 2s → 4s)
Pagination	Use page and pageSize parameters where available
Error Handling	Check the code field for success or retryable errors
Resource Lifecycle	Always pause or terminate unused Pods to save cost

Tip: Bundle related API calls in scripts to minimize overhead. Efficient API usage helps lower latency and improve cluster utilization.

---

💬 Support & Feedback

We’re always improving the Yotta API. For questions, feature requests, or bug reports, contact:

📩 [email protected] 🌐 Yotta Labs Official Website 🧑‍💻 Yotta Platform Console