Try Chat Free Docs Pricing AI Analysis Repo Insights Free Tools About Blog
Sign in Get started free
< Back to blog
Tutorial Velona Team ·16 September 2026 ·6 min read

How to Use Velona with OpenAI-Compatible Coding Agents

Use Velona with Your AI Coding Agent

AI coding agents are changing how developers build software. Tools like OpenCode, Zed, Cline, Continue, and other agentic coding platforms can inspect your codebase, read files, edit code, run commands, and work through multi-step development tasks.

Many of these tools support OpenAI-compatible API endpoints. This means you can use Velona as the AI gateway behind your coding agent without building a custom integration, and pay for it in rupees with UPI from ₹10.

In this guide, we'll show you how to connect Velona to an OpenAI-compatible coding platform in just a few steps, and which models give you the best value.

What You Need

You only need three things:

Velona's OpenAI-compatible API uses the following base URL:

https://velona.in/v1

Your coding agent will use this endpoint just like it would use any other OpenAI-compatible provider.

Which Model Should Your Coding Agent Use?

Coding agents resend your code as input on every step, so the input price matters most. One step with about 30,000 input tokens and 1,000 output tokens costs, at Velona's prices on 4 October 2026:

ModelModel IDCost per agent stepWhen to use it
DeepSeek V4 Flashdeepseek/deepseek-v4-flashabout ₹0.09Default for everyday coding
DeepSeek V4 Prodeepseek/deepseek-v4-proabout ₹0.69When Flash gets stuck
Qwen3 Coderqwen/qwen3-coderabout ₹1.04An alternative coding specialist
Kimi K2.6moonshotai/kimi-k2.6about ₹3.37Long agentic runs
Claude Sonnet 5anthropic/claude-sonnet-5about ₹7.25The hardest bugs and planning

A good setup is a cheap model for doing the work and a strong model only for planning or tricky problems. Our guide to the cheapest models for coding agents compares 13 models on real rupee costs per task.

How the Connection Works

The basic architecture is simple:

AI Coding Agent
      |
      | OpenAI-compatible API
      v
https://velona.in/v1
      |
      v
    Velona
      |
      v
 AI Model

Your coding agent handles the development workflow and tools, while Velona provides the model through its API.

You don't need a separate Velona SDK for this setup. If your coding platform allows a custom OpenAI-compatible endpoint, you can point it to Velona.

Step 1: Get Your Velona API Key

Sign in to Velona and create an API key on the API Keys page. Give it a budget so a runaway agent loop can never drain your wallet.

Your API key is a secret credential, so keep it private. Do not commit it to Git repositories or expose it in frontend applications.

For local development, you can store the key in an environment variable:

VELONA_API_KEY=your_api_key

Step 2: Find an Available Model

Your coding agent needs a model ID to send requests to Velona. The table above lists good choices for coding, and every model ID is in the model catalogue.

You can also retrieve the available models using the OpenAI-compatible models endpoint:

curl https://velona.in/v1/models \
  -H "Authorization: Bearer YOUR_VELONA_API_KEY"

Use the exact model ID returned by Velona when configuring your coding agent.

Step 3: Configure Your Coding Agent

Most OpenAI-compatible coding platforms ask for the same basic information:

Provider: Velona

Base URL:
https://velona.in/v1

API Key:
YOUR_VELONA_API_KEY

Model:
deepseek/deepseek-v4-flash

Once these values are configured, the coding agent can communicate with Velona through the OpenAI-compatible API.

Using Velona with OpenCode

OpenCode supports custom AI providers, including OpenAI-compatible providers.

Configure Velona as a custom provider and use:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "velona": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Velona",
      "options": {
        "baseURL": "https://velona.in/v1"
      },
      "models": {
        "deepseek/deepseek-v4-flash": { "name": "DeepSeek V4 Flash" },
        "deepseek/deepseek-v4-pro": { "name": "DeepSeek V4 Pro" },
        "anthropic/claude-sonnet-5": { "name": "Claude Sonnet 5" }
      }
    }
  }
}

Authenticate OpenCode with your Velona API key and select the Velona provider and model.

Using Velona with Cline

In Cline's settings, choose OpenAI Compatible as the API provider, set the Base URL to https://velona.in/v1, paste your Velona API key, and enter a model ID such as deepseek/deepseek-v4-flash.

Using Velona with Zed

Zed supports custom OpenAI-compatible providers.

In Zed's Agent or AI provider settings, add a custom OpenAI-compatible provider and enter:

Provider:
Velona

API URL:
https://velona.in/v1

API Key:
YOUR_VELONA_API_KEY

Model:
deepseek/deepseek-v4-flash

Save the configuration and select the Velona model from the Agent interface.

Using Velona with Other Coding Agents

The same approach works with other coding platforms that allow you to configure a custom OpenAI-compatible endpoint, such as Continue and Roo Code.

Look for settings such as:

Set the API URL to:

https://velona.in/v1

Then enter your Velona API key and select an available model.

The exact configuration will vary between coding platforms, but the underlying connection remains the same.

Why Use Velona for Agentic Coding?

A coding agent is more than a chatbot. It can perform a sequence of actions while working on a software project.

A typical agentic coding workflow may look like this:

Developer
    |
    v
Coding Agent
    |
    +-- Read files
    |
    +-- Search code
    |
    +-- Understand the task
    |
    +-- Ask the AI model
    |
    +-- Use tools
    |
    +-- Edit files
    |
    +-- Run tests
    |
    +-- Inspect results
    |
    +-- Make additional changes
    |
    v
Completed task

Velona sits behind this workflow and provides the AI model through a standard API interface.

This separation is useful because you can change the model or model provider without having to change your coding workflow. You also see the rupee cost of every step on your usage page.

Test Your Connection

After configuring your coding agent, start with a simple task.

For example:

Create a simple Python Hello World program.

Then test repository understanding:

Inspect this project and explain its architecture.
Do not modify any files.

Finally, test an agentic workflow:

Find where authentication is implemented in this project.

Explain which files need to change to add a logout endpoint.

Do not modify anything yet.

If your agent can successfully inspect the project and use its available tools while communicating with the model through Velona, your OpenAI-compatible integration is working as expected.

Keep Your API Key Secure

Your Velona API key should be treated like a password.

Avoid putting it directly into source code:

// Avoid this

apiKey: "your-secret-api-key"

Instead, use environment variables or the secure credential storage provided by your coding platform.

If you use a .env file locally, make sure it is excluded from Git:

.env

One Endpoint for Your Coding Workflow

The main advantage of an OpenAI-compatible interface is simplicity.

Your coding agent only needs to know:

Base URL
API Key
Model ID

The workflow becomes:

OpenCode / Zed / Coding Agent
            |
            v
    Velona OpenAI API
    https://velona.in/v1
            |
            v
       AI Model

Instead of configuring a different integration for every model provider, you can use Velona as the API layer between your coding environment and the models you want to use.

Start Coding with Velona

If your AI coding platform supports custom OpenAI-compatible endpoints, connecting it to Velona is straightforward.

Get your Velona API key, choose an available model, set the base URL to https://velona.in/v1, and start coding.

One API endpoint. Multiple models. Your existing coding workflow. Billed in rupees.

Get started with Velona and top up from ₹10 with UPI.