First Request Example

Run a minimal working call from model listing to your first Chat Completions request.

This page shows the shortest practical call flow: verify your API key first, then send your first model request. The examples use the standard OpenAI-compatible entry point, which is the recommended starting point for new integrations, backend services, and command-line debugging.

Before you start, prepare:

Item	Example	Notes
API Base URL	https://api.nusabasedigital.com	The English docs use this domain in examples.
API Key	sk-your_api_key	Send it as a Bearer token in the request header.
Model	qwen3.5-flash	This example model appears in the current pricing docs. Use GET /v1/models as the source of truth for your account.

If you are not sure which header to use, read Authentication first.

1. Set Environment Variables

Put the base URL and API key into environment variables so the following examples can be copied directly.

export NUSABASE_BASE_URL="https://api.nusabasedigital.com"
export NUSABASE_API_KEY="sk-your_api_key"

Windows PowerShell:

$env:NUSABASE_BASE_URL = "https://api.nusabasedigital.com"
$env:NUSABASE_API_KEY = "sk-your_api_key"

2. List Models

Start with GET /v1/models. This does not generate model output, so it is the safest first call for checking your API key, account status, and available models.

curl "$NUSABASE_BASE_URL/v1/models" \
  -H "Authorization: Bearer $NUSABASE_API_KEY"

If the request succeeds, the response contains the models your account can access. Choose the model value in later requests from this response.

You can also retrieve one model:

curl "$NUSABASE_BASE_URL/v1/models/qwen3.5-flash" \
  -H "Authorization: Bearer $NUSABASE_API_KEY"

If this step fails, start with these checks:

Status	Common cause	What to check
401	Missing, invalid, expired, or exhausted API key	Check Authorization: Bearer ... and make sure you are using an API key.
403	Account, group, IP allowlist, or API access permission is not valid for the request	Check account status, key permissions, group access, and IP restrictions.
503	The selected model or channel is temporarily unavailable	Retry later or use another model returned by the model list.

3. Send Your First Chat Request

Use POST /v1/chat/completions for the first inference request. It follows the OpenAI Chat Completions shape and has a small, easy-to-debug request body.

curl "$NUSABASE_BASE_URL/v1/chat/completions" \
  -H "Authorization: Bearer $NUSABASE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5-flash",
    "messages": [
      {
        "role": "system",
        "content": "You are a concise and accurate assistant."
      },
      {
        "role": "user",
        "content": "Introduce Nusabase API in one sentence."
      }
    ]
  }'

Request body fields:

Field	Required	Description
model	Yes	The model to call. Prefer a model returned by GET /v1/models.
messages	Yes	Conversation message array, usually with system, user, and assistant roles.
role	Yes	Message role. The first user input should use user.
content	Yes	Message content. Text models can use a plain string.

A successful response usually contains choices; the model output is in choices[0].message.content.

{
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "Nusabase API is an aggregation platform for calling multiple AI models through compatible APIs."
      }
    }
  ]
}

The sample response only shows the key fields. Real responses may also include id, object, created, model, usage, and other fields.

4. Enable Streaming

Set stream to true if you want to receive tokens as they are generated.

curl "$NUSABASE_BASE_URL/v1/chat/completions" \
  -H "Authorization: Bearer $NUSABASE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5-flash",
    "stream": true,
    "messages": [
      {
        "role": "user",
        "content": "Give me a three-point summary."
      }
    ]
  }'

Streaming is useful for chat UIs, long generation tasks, and any workflow that should show the first output quickly. Your client should read the response incrementally using Server-Sent Events or a compatible streaming parser.

5. Responses Example

If your client already targets the OpenAI Responses style, call POST /v1/responses.

curl "$NUSABASE_BASE_URL/v1/responses" \
  -H "Authorization: Bearer $NUSABASE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5-flash",
    "input": "Introduce Nusabase API in one sentence."
  }'

The project also exposes a context-compaction endpoint:

curl "$NUSABASE_BASE_URL/v1/responses/compact" \
  -H "Authorization: Bearer $NUSABASE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5-flash",
    "input": "Compact this conversation context."
  }'

Whether a model supports Responses or compaction depends on the available backend channel configuration. If the request returns a model or channel availability error, switch to a model returned by GET /v1/models.

6. Compatibility Examples

If you already have a Claude or Gemini SDK, you can use the compatible entry points for a first smoke test.

Claude Messages example:

curl "$NUSABASE_BASE_URL/v1/messages" \
  -H "x-api-key: $NUSABASE_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5-flash",
    "max_tokens": 128,
    "messages": [
      {
        "role": "user",
        "content": "hello"
      }
    ]
  }'

Gemini model-list example:

curl "$NUSABASE_BASE_URL/v1beta/models" \
  -H "x-goog-api-key: $NUSABASE_API_KEY"

You can also use the query-string key form:

curl "$NUSABASE_BASE_URL/v1beta/models?key=$NUSABASE_API_KEY"

Compatibility entry points may use different authentication headers and response shapes from the standard /v1/* API. See Compatibility Formats for details.

7. Next Steps

After your first request succeeds, continue with:

Doc	Use it for
Authentication	API key, Bearer auth, Claude, Gemini, and WebSocket authentication.
Models and Pricing	Choosing a model and checking input, output, cache, and other prices.
Billing Rules	Understanding usage-based, per-call, duration-based, resolution-based, and tiered billing.
Error Codes	Debugging failures by HTTP status and error.code.

References

• Authentication
• Compatibility Formats
• Models and Pricing
• Error Codes