> ## Documentation Index
> Fetch the complete documentation index at: https://docs.neosantara.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Penalaran Responses API

> Kontrol kedalaman pemikiran model menggunakan OpenAI SDK dan parameter reasoning effort.

Pada endpoint `/v1/responses`, Anda dapat mengatur intensitas proses penalaran model AI menggunakan parameter `reasoning` melalui [OpenAI SDK](https://openai.com/?utm_source=neosantara-docs\&utm_medium=referral) resmi.

## Konfigurasi Tingkat Intensitas Penalaran

Gunakan parameter `reasoning` (atau `extra_body={"reasoning": ...}` pada Python SDK) untuk menentukan anggaran penalaran model.

<CodeGroup>
  ```python Python (OpenAI SDK) icon="python" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  from openai import OpenAI
  import os

  client = OpenAI(
      base_url="https://api.neosantara.xyz/v1",
      api_key=os.environ["NEOSANTARA_API_KEY"]
  )

  # Gunakan parameter reasoning secara langsung (atau extra_body pada SDK versi lama)
  response = client.responses.create(
      model="deepseek-v4.1-flash",
      input="Buktikan secara matematis mengapa akar kuadrat dari 2 adalah bilangan irasional.",
      reasoning={
          "effort": "high"
      }
  )

  print("Hasil Analisis:\n", response.output_text)
  ```

  ```javascript TypeScript (Node.js) icon="js" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  import OpenAI from "openai";

  const client = new OpenAI({
    baseURL: "https://api.neosantara.xyz/v1",
    apiKey: process.env.NEOSANTARA_API_KEY,
  });

  const response = await client.responses.create({
    model: "deepseek-v4.1-flash",
    input: "Buktikan secara matematis mengapa akar kuadrat dari 2 adalah bilangan irasional.",
    // @ts-expect-error - ekstensi skema responses
    reasoning: {
      effort: "high",
    },
  });

  console.log("Hasil Analisis:\n", response.output_text);
  ```
</CodeGroup>

## Mengakses Jejak Pemikiran dan Akuntansi Token

Saat menggunakan model penalaran seperti `deepseek-r1`, Anda dapat memeriksa alokasi token penalaran internal melalui properti `usage.completion_tokens_details`.

<CodeGroup>
  ```python Python (OpenAI SDK) icon="python" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  from openai import OpenAI
  import os

  client = OpenAI(
      base_url="https://api.neosantara.xyz/v1",
      api_key=os.environ["NEOSANTARA_API_KEY"]
  )

  response = client.responses.create(
      model="deepseek-r1",
      input="Berapa banyak digit nol di akhir dari 100 faktorial (100!)?",
      reasoning={"effort": "high"}
  )

  # Memeriksa rincian token penalaran
  print(f"Token Output: {response.usage.completion_tokens}")
  if hasattr(response.usage, "completion_tokens_details"):
      details = response.usage.completion_tokens_details
      print(f"Token Penalaran: {getattr(details, 'reasoning_tokens', 'N/A')}")

  print("\nJawaban Final:\n", response.output_text)
  ```

  ```javascript TypeScript (Node.js) icon="js" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  import OpenAI from "openai";

  const client = new OpenAI({
    baseURL: "https://api.neosantara.xyz/v1",
    apiKey: process.env.NEOSANTARA_API_KEY,
  });

  const response = await client.responses.create({
    model: "deepseek-r1",
    input: "Berapa banyak digit nol di akhir dari 100 faktorial (100!)?",
    // @ts-expect-error - ekstensi skema responses
    reasoning: { effort: "high" },
  });

  console.log(`Token Output: ${response.usage?.completion_tokens}`);
  // @ts-expect-error - reasoning_tokens property
  console.log(`Token Penalaran: ${response.usage?.completion_tokens_details?.reasoning_tokens}`);
  console.log("\nJawaban Final:\n", response.output_text);
  ```
</CodeGroup>

## Tingkat Intensitas `effort`

| Nilai      | Keterangan                                                                                             |
| :--------- | :----------------------------------------------------------------------------------------------------- |
| `"low"`    | Alokasi anggaran penalaran minimal. Cocok untuk tugas terstruktur yang membutuhkan respons cepat.      |
| `"medium"` | Keseimbangan optimal antara kecepatan dan kedalaman pemikiran (default).                               |
| `"high"`   | Alokasi penalaran maksimum untuk pembuktian matematika, analisis kode rumit, dan logika multi-langkah. |

## Langkah Selanjutnya

| Kebutuhan                      | Panduan                                                                |
| :----------------------------- | :--------------------------------------------------------------------- |
| Extended Thinking di Anthropic | [Anthropic Extended Thinking](/id/gateway/anthropic-messages/thinking) |
| Reasoning di Chat Completions  | [Penalaran OpenAI](/id/gateway/chat-completions/reasoning)             |
| Tugas Latar Belakang           | [Background Tasks](/id/gateway/responses-api/background-jobs)          |


## Related topics

- [OpenResponses API](/id/gateway/responses-api.md)
- [Penalaran (Reasoning)](/id/gateway/chat-completions/reasoning.md)
- [Extended Thinking Anthropic](/id/gateway/anthropic-messages/thinking.md)