Test your AI application with Divyam

Connect your AI application to the Divyam demo endpoint, send test requests, and review their results in the console. You will create an API key, configure the connection, run a first request, and test your application’s workflows.

Start with the Self-serve UI guide for account setup and console navigation. Developers can use the full Python, Node.js, or HTTP integration guide.

For a coding harness, use the separate Pi setup guide. For browser-based chat testing, use the Open WebUI guide.

Before you start

You need access to your email and permission to change your application’s model connection. The connection uses OpenAI Chat Completions. The Python first-request example requires Python and the OpenAI SDK.

Copy the base URL and model ID from your console. Keep the complete URL path and model prefix.

1 Sign up or sign in

1. Open the Divyam demo console.
2. Choose signup for a new account, or sign in if you already have one.
3. Complete email verification using the code sent to your inbox.
4. Open Getting Started.
5. Check your available credit in Credits & Billing.

If access is restricted, contact the demo administrator with your email address. Use the credit and account limits shown in your console.

2 Create an API key

1. Select API Keys in the left navigation.
2. Create a key with a recognizable name, such as my-app-demo.
3. Copy the complete key when it appears.
4. Store it privately so you can configure your application’s connection.

The key is shown once. If you lose it, create a replacement. Use this API key for application requests. The email verification code and browser session token are separate credentials.

Keep the key in your backend or trusted connection settings. Do not put it in browser JavaScript or commit it.

3 Copy the base URL and choose a model

Open Getting Started and copy the API base URL from the example request. Then open Models & Pricing and copy an available model ID exactly.

The demo currently lists divyam-as/gpt-5.6-sol and divyam-as/gpt-6-sol. The examples below use divyam-as/gpt-5.6-sol. Substitute your chosen model consistently in the configuration and requests.

Choosing a model in Getting Started changes the example shown there. Your application must send its chosen model ID with each request. That requested catalog model determines the applicable billing rates.

4 Send your first request

Run this request to check your key, base URL, and model before testing the application’s full workflow.

Install the Python SDK

Run this in your application’s Python environment:

python3 -m pip install openai

Set the connection values

In the terminal where you will run the example, set the base URL and model:

export DIVYAM_BASE_URL='https://api.demo.divyam.ai/v1'
export DIVYAM_MODEL='divyam-as/gpt-5.6-sol'

Run these commands in Bash. At the prompt, paste the key from step 2 and press Enter. The key stays hidden while you enter it.

read -rsp 'Divyam API key: ' DIVYAM_API_KEY
printf '\n'
export DIVYAM_API_KEY

read sets DIVYAM_API_KEY. export makes it available to the application launched from that terminal. Keep using the same terminal. A new terminal needs these variables set again.

Create and run the request

Save the following as first_request.py:

"""Send one test message through Divyam."""

import os

from openai import OpenAI

client = OpenAI(
    base_url=os.environ["DIVYAM_BASE_URL"],
    api_key=os.environ["DIVYAM_API_KEY"],
)
response = client.chat.completions.create(
    model=os.environ["DIVYAM_MODEL"],
    messages=[{"role": "user", "content": "Say hello in one sentence."}],
)
print(response.choices[0].message.content)

Run it:

python3 first_request.py

You should see an answer in the terminal. Note the request time so you can find it in Divyam Logs. If the request fails, resolve the error using the troubleshooting section before continuing.

5 Configure your application

Use the same key, base URL, and model that worked in the first request.

Applications with connection settings

Open the application’s settings and select a custom provider or OpenAI-compatible connection. The field labels vary between applications.

Save the connection and select it for the workflow you want to test. For an application managed by someone else, ask its administrator to configure these values.

Applications with a custom backend

Use the client configuration from step 4 in the part of your backend that makes model requests. Keep your application’s messages and workflow logic. Set the request’s model to your chosen Divyam catalog ID.

Store the connection values in your backend’s existing configuration or secret settings. If the application is already running, restart or reload it as needed to apply the new settings.

Check the URL format

A base-URL field should contain https://api.demo.divyam.ai/v1 for this demo. If the application asks for the complete completion URL, use https://api.demo.divyam.ai/v1/chat/completions.

Do not append /chat/completions to a field that adds it automatically. This guide uses Chat Completions. Applications requiring Responses or native Anthropic Messages need a separate compatibility check.

6 Test your application’s workflows

Run representative inputs through the application itself. A successful first request checks connectivity; these tests check whether your application behaves as expected.

1. Run a typical task and inspect the application’s answer.
2. Run a more demanding input from your existing test cases.
3. If the application maintains conversations, send a follow-up and check that it uses the earlier context.
4. If it uses tools, verify the actual tool action and the application’s response after the tool result.
5. Record the input, expected behavior, actual result, requested model, and test time.

Use your existing acceptance criteria to assess the answers. Check any enabled features, such as streaming or image input, through their normal application workflow.

Repeat the same test cases when comparing available Divyam models. Update the application configuration and requested model together so each run uses the intended connection.

7 Review the requests in Divyam

1. Return to the Divyam demo console. Sign in again if needed.
2. Select Logs in the left navigation.
3. Set Model to the ID used by your application.
4. Set Period to include your test. Leave Status at All statuses to include failures.
5. Find requests matching your test times. The newest requests appear first.
6. Check the outcome, token counts, and latency. Successful requests should show a 2xx status.
7. Open View request to confirm the model and input when request content is available.
8. Open View response when available to inspect the recorded answer.
9. Open Usage to review request totals, token usage, and spend.
10. Open Credits & Billing to review settled charges and remaining credit.

The demo displays request times in UTC. Convert your local test time when comparing timestamps. If recent requests are missing, check the account and filters, then refresh after records arrive.

For streamed responses, the console may show Streamed; not stored instead of View response. For tool workflows, a continuation request can contain the earlier assistant tool call and its matching tool result. Match the tool-call ID when checking that exchange.

Divyam’s completion model field identifies the requested catalog model. It does not identify the backing model that answered. Use a separate requested-versus-routed view when available.

Your test is complete when the application produces the expected behavior and its requests appear under the intended catalog model. Review billing after records settle before drawing conclusions about the cost of the workflow.

Troubleshooting

Rotate a key or restore your previous connection

To rotate a key, create a replacement, update the application, verify one request, then revoke the old key. Follow the API key rotation guide for the console steps and key-limit checks. To end the demo, restore the application’s previous connection and revoke a dedicated demo key when no longer needed.